Methods and systems for non-invasive prediction of prognosis impairment in IgA nephropathy, determination of response to immunosuppressive treatment

Abstract

The disclosure provides an IgA nephropathy prognosis injury non-invasive prediction and immunosuppressive treatment response determination method and system, relating to the technical field of IgA nephropathy prognosis injury non-invasive prediction and immunosuppressive treatment response. The IgA nephropathy prognosis injury non-invasive prediction method comprises: acquiring preset peripheral blood indexes corresponding to a cross-sectional pathological prediction cohort and corresponding IgA nephropathy prognosis pathological injury labels, and preset peripheral blood indexes to be predicted; constructing a corresponding prognosis injury non-invasive prediction model according to the preset peripheral blood indexes and the IgA nephropathy prognosis pathological injury labels; and performing prognosis injury prediction based on the prognosis injury non-invasive prediction model and the preset peripheral blood indexes to be predicted. The embodiment of the disclosure can realize IgA nephropathy prognosis injury non-invasive prediction and immunosuppressive treatment response determination.

Description

Technical Field

[0001] This disclosure relates to the field of non-invasive prediction of prognostic damage in IgA nephropathy and response to immunosuppressive therapy, and particularly to a method and system for non-invasive prediction of prognostic damage in IgA nephropathy and determination of response to immunosuppressive therapy. Background Technology

[0002] IgA nephropathy (immunoglobulin A nephropathy) is the most common primary glomerular disease worldwide, especially among Asian populations, and remains the most common cause of chronic kidney disease. Primary IgA nephropathy presents with diverse clinical and pathological manifestations and can occur at any age, but it is most prevalent in young adults aged 30-40. It is one of the leading causes of end-stage renal disease (ESRD). Recent studies show that within 20-25 years, 15-30% of IgA nephropathy patients will eventually progress to ESRD, requiring dialysis or kidney transplantation. IgA nephropathy accounts for 45.3-54.3% of all primary glomerular diseases. As the most common primary glomerular disease, IgA nephropathy has a relatively poor prognosis, placing a significant burden on individuals and society. Because it commonly affects young adults, it significantly reduces the workforce.

[0003] The significance of non-invasive prediction of prognostic damage in IgA nephropathy (IgAN) lies primarily in four core dimensions: compensating for the limitations of invasive examinations, achieving early warning and accurate prognostic assessment, guiding individualized clinical decision-making, and promoting precision disease management. First, it reduces reliance on renal biopsies: While renal biopsy is the "gold standard" for diagnosing and assessing the prognosis of IgAN, it is an invasive procedure with risks of complications such as bleeding and infection, and is not suitable for repeated procedures. Non-invasive prediction (such as urine / blood biomarker testing) can replace or supplement biopsies, enabling repeated, dynamic monitoring of the condition and reducing the medical risks and psychological burden on patients. Second, it enables early and dynamic monitoring: Non-invasive tests (such as urine samples) are readily available and can be repeatedly collected in the early stages of the disease or during follow-up, thereby dynamically monitoring the progression of kidney damage and overcoming the limitation that biopsies can only reflect the pathological state at a "specific point in time." Second, early identification of high-risk patients: The prognosis of IgAN varies greatly, and some patients have insidious disease. Non-invasive predictive models (combining clinical indicators and multi-omics biomarkers) can identify high-risk patients with poor prognosis (such as rapid progression to end-stage renal disease) at an early stage, achieving "early detection and early intervention." Precise quantification of damage extent: Non-invasive prediction can quantitatively assess microscopic damage such as the area of ​​renal tubulointerstitial fibrosis (IF) and the degree of inflammation, providing a more accurate prognostic assessment than traditional single clinical indicators (such as urine protein and serum creatinine), compensating for the insensitivity of traditional indicators to early or mild damage. Third, assistance in treatment decision-making: Non-invasive prediction results can provide clinicians with objective quantitative evidence to help determine whether patients need more aggressive intervention (such as early use of immunosuppressants) or adjustments to the treatment plan, achieving "precision medicine." Assessment of treatment response: Through the dynamic changes of non-invasive indicators, the early response of patients to specific treatments (such as glucocorticoids) can be non-invasively assessed, avoiding blind medication and achieving dynamic management of the treatment process. Fourth, elucidating disease mechanisms: the discovery of non-invasive predictive biomarkers (such as specific inflammatory factors, complement components, and urinary proteins) helps to deepen the understanding of the mechanisms of IgAN progression and kidney injury, providing a basis for the development of new targeted therapies; optimizing disease management: non-invasive prediction helps to establish individualized risk stratification models for IgAN, providing data support for clinical research, disease cohort studies, and the rational allocation of medical resources, ultimately improving patients' long-term kidney survival.

[0004] Meanwhile, the prognosis of IgAN is closely related to the response to immunosuppressive therapy (such as glucocorticoids and immunosuppressants). Assessing treatment response is not only about short-term symptom relief, but also directly determines the patient's long-term renal prognosis and individualized treatment decisions.

[0005] Therefore, it is necessary to propose a technical method for non-invasive prediction of prognostic damage in IgA nephropathy and corresponding response to immunosuppressive therapy, in order to solve at least one of the technical problems currently faced, such as the inability to perform non-invasive prediction of prognostic damage in IgA nephropathy, the inability to improve the performance of non-invasive prediction models for prognostic damage in IgA nephropathy, and the inability to effectively determine the response to immunosuppressive therapy. Summary of the Invention

[0006] This disclosure presents a technical solution for a non-invasive prediction of prognostic damage in IgA nephropathy and a method and system for determining the response to immunosuppressive therapy.

[0007] According to one aspect of this disclosure, a non-invasive method for predicting prognostic damage in IgA nephropathy is provided, comprising: acquiring preset peripheral blood indicators corresponding to a cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, and preset peripheral blood indicators to be predicted; constructing a corresponding non-invasive prognostic damage prediction model based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels; and predicting prognostic damage based on the non-invasive prognostic damage prediction model and the preset peripheral blood indicators to be predicted.

[0008] Preferably, determining the preset peripheral blood index includes: acquiring multiple peripheral blood indices to be processed corresponding to the cross-sectional pathological prediction cohort; performing logistic regression analysis on the multiple peripheral blood indices to be processed and their corresponding IgA nephropathy prognostic pathological damage labels to obtain multiple levels of significance; configuring the peripheral blood indices to be processed corresponding to the multiple levels of significance less than the preset level of significance as preset peripheral blood indices; and / or, configuring the prognostic damage and the IgA nephropathy prognostic pathological damage labels to reflect renal tubular atrophy and interstitial fibrosis corresponding to chronic irreversible damage or glomerular endothelial cell hyperplasia corresponding to acute active inflammation.

[0009] Preferably, the IgA nephropathy is configured as IgA nephropathy induced by tumor lesions; wherein, the prediction of IgA nephropathy induced by tumor lesions includes: training a preset predictor using quantitative imaging features of preset tumor lesions corresponding to tumor lesions in medical images of preset locations at a first time point and pathological results corresponding to induced or non-induced IgA nephropathy at a second time point after the first time point to construct a prediction model for IgA nephropathy induced by tumor lesions; and using the prediction model for IgA nephropathy induced by tumor lesions and the quantitative imaging features of preset tumor lesions corresponding to tumor lesions in medical images of preset locations to be predicted to predict IgA nephropathy induced by tumor lesions.

[0010] Preferably, before training the preset predictor, the method includes: extracting quantitative imaging features of tumor lesions corresponding to tumor lesions in medical images of preset locations; and selecting preset quantitative imaging features of tumor lesions that are related to the pathological results from the quantitative imaging features of the tumor lesions to be selected.

[0011] Preferably, determining the preset quantitative imaging features of tumor lesions related to the pathological results includes: using a univariate logistic regression analysis algorithm, the quantitative imaging features of the selected tumor lesions and their corresponding pathological results, performing univariate logistic regression on each of the quantitative imaging features of the selected tumor lesions to determine the preset quantitative imaging features of tumor lesions related to the pathological results.

[0012] Preferably, the quantitative imaging features of the selected tumor lesion include one or more of the following: the long diameter of the tumor lesion, the long diameter of the solid component of the tumor lesion, the ratio of the solid component of the tumor lesion to the total tumor lesion, the location of the tumor lesion, whether the edge of the tumor lesion is clear, the morphology of the tumor lesion, the spinous process sign of the tumor lesion, the lobulation sign of the tumor lesion, the pleural traction sign of the tumor lesion, the cavitation sign of the tumor lesion, and the air bronchus sign of the tumor lesion.

[0013] Preferably, determining whether the edge of the tumor lesion corresponding to the quantitative imaging features of the selected tumor lesion is clear includes: performing an erosion operation on the tumor lesion mask image to obtain a tumor lesion mask erosion image; subtracting the tumor lesion mask erosion image from the tumor lesion mask image to obtain a tumor lesion mask processed image; mapping the tumor lesion location coordinates in the tumor lesion mask processed image to the preset site medical image; calculating multiple gradients corresponding to the tumor lesion location coordinates in the preset site medical image; if the gradient magnitude values ​​corresponding to the multiple tumor lesions are greater than or equal to a preset gradient magnitude value, then the boundary of the tumor lesion location corresponding to the tumor lesion location greater than or equal to the preset gradient magnitude value is determined to be clear; otherwise, the tumor lesion boundary is determined to be unclear.

[0014] Preferably, determining the spur includes: detecting the edges of the tumor lesion mask in the tumor lesion mask image to obtain the tumor lesion mask edge line; calculating the first pixel distance and the second pixel distance between any tumor lesion mask edge pixel on the tumor lesion mask edge line and its adjacent tumor lesion mask edge pixels; if the first pixel distance and the second pixel distance are greater than a set pixel distance, then configuring the arbitrary tumor lesion mask edge pixel as a spur; and / or, determining the lobulation sign of the tumor lesion includes: performing an erosion operation on the tumor lesion mask image corresponding to each tumor lesion in the medical image of the preset location to obtain a tumor lesion mask erosion image; subtracting the corresponding tumor lesion mask erosion image from the tumor lesion mask image of each tumor lesion to obtain a tumor lesion mask processing image corresponding to each tumor lesion; judging the morphology of the tumor lesion in the tumor lesion mask processing image corresponding to each tumor lesion; if the morphology of the tumor lesion is irregular, then determining whether there is a tumor lesion lobulation sign in the tumor lesion mask processing image corresponding to each tumor lesion.

[0015] According to one aspect of this disclosure, a method for determining the response to immunosuppressive therapy is provided, comprising: using a non-invasive prediction method for prognostic damage in IgA nephropathy as described above, determining a longitudinal treatment follow-up cohort corresponding to prognostic damage in the cross-sectional pathological prediction cohort; and using an estimated glomerular filtration rate as an outcome, determining the differences in response to immunosuppressive therapy under different risk stratifications of endothelial cell proliferation.

[0016] According to one aspect of this disclosure, a non-invasive prediction system for prognostic damage in IgA nephropathy is provided, comprising: an acquisition unit for acquiring preset peripheral blood indicators corresponding to a cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, and preset peripheral blood indicators to be predicted; a construction unit for constructing a corresponding non-invasive prediction model for prognostic damage based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels; and a prediction unit for predicting prognostic damage based on the non-invasive prediction model for prognostic damage and the preset peripheral blood indicators to be predicted; or, comprising: a processor for storage and processing. The processor may include: a memory containing executable instructions; wherein the processor is configured to invoke instructions stored in the memory to execute the aforementioned non-invasive prediction method for prognostic damage in IgA nephropathy; or, a computer-readable storage medium storing a computer program / instructions and a bitstream thereon, wherein the computer program / instructions, when executed by a processor, implement the aforementioned non-invasive prediction method for prognostic damage in IgA nephropathy to generate the bitstream; or, a computer program product configured with a computer program / instructions, wherein the computer program / instructions, when executed by a processor, implement the aforementioned non-invasive prediction method for prognostic damage in IgA nephropathy.

[0017] According to one aspect of this disclosure, an immunosuppressive therapy response determination system is provided, comprising: an acquisition unit for acquiring preset peripheral blood indicators corresponding to a cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, and preset peripheral blood indicators to be predicted; a construction unit for constructing a corresponding non-invasive prognostic damage prediction model based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels; a first determination unit for determining a longitudinal treatment follow-up cohort corresponding to prognostic damage in the cross-sectional pathological prediction cohort based on the non-invasive prognostic damage prediction model and the preset peripheral blood indicators to be predicted; and a second determination unit for determining different outcomes based on an estimated glomerular filtration rate. Differences in response to immunosuppressive therapy under risk stratification of endothelial cell proliferation; or, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke instructions stored in the memory to perform the above-described immunosuppressive therapy response determination method; or, comprising: a computer-readable storage medium having stored thereon a computer program / instructions and a bit stream thereon, wherein the computer program / instructions, when executed by a processor, implement the above-described immunosuppressive therapy response determination method to generate the bit stream; or, comprising: a computer program product having configured computer programs / instructions, wherein the computer program / instructions, when executed by a processor, implement the above-described immunosuppressive therapy response determination method.

[0018] In this disclosure, a technical solution is proposed for a non-invasive prediction of prognostic damage in IgA nephropathy and a method and system for determining the response to immunosuppressive therapy. This addresses at least one of the following technical problems: the inability to perform non-invasive prediction of prognostic damage in IgA nephropathy, the inability to improve the performance of non-invasive prediction models for IgA nephropathy prognostic damage, and the inability to effectively determine the response to immunosuppressive therapy.

[0019] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Other features and aspects of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the specification, serve to illustrate the technical solutions of this disclosure.

[0021] Figure 1 A flowchart illustrating a non-invasive method for predicting prognostic damage in IgA nephropathy according to an embodiment of the present disclosure is shown. Figure 2 The diagram shows the visualization results corresponding to the embodiments of this disclosure. Detailed Implementation

[0022] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. Like reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise. The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0023] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0024] Furthermore, to better illustrate this disclosure, numerous specific details are provided in the following detailed embodiments. Those skilled in the art should understand that this disclosure can be practiced even without certain specific details. In some instances, methods, means, components, and circuits well-known to those skilled in the art have not been described in detail to highlight the main points of this disclosure. It is understood that the various method embodiments mentioned above in this disclosure can be combined with each other to form combined embodiments without violating the underlying principles and logic; due to space limitations, these will not be elaborated upon further.

[0025] In addition, this disclosure also provides a device or system for non-invasive prediction of prognostic damage in IgA nephropathy and determination of response to immunosuppressive therapy, electronic devices, computer-readable storage media, and program products. All of the above can be used to implement any of the methods for non-invasive prediction of prognostic damage in IgA nephropathy and determination of response to immunosuppressive therapy provided in this disclosure. The corresponding technical solutions and descriptions are described in the relevant section on methods for non-invasive prediction of prognostic damage in IgA nephropathy and determination of response to immunosuppressive therapy, and will not be repeated here.

[0026] Figure 1 A flowchart illustrating a non-invasive method for predicting prognostic damage in IgA nephropathy according to embodiments of the present disclosure is shown, such as... Figure 1As shown, the non-invasive prediction method for prognostic damage in IgA nephropathy includes: Step S1: Obtaining preset peripheral blood indicators corresponding to the cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, as well as preset peripheral blood indicators to be predicted; Step S2: Constructing a corresponding non-invasive prediction model for prognostic damage based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels; Step S3: Predicting prognostic damage based on the non-invasive prediction model for prognostic damage and the preset peripheral blood indicators to be predicted. This method aims to solve at least one of the following technical problems: the inability to perform non-invasive prediction of prognostic damage in IgA nephropathy, improving the performance of non-invasive prediction models for IgA nephropathy prognostic damage, and the inability to effectively determine the response to immunosuppressive therapy.

[0027] Specifically, the step of constructing a corresponding non-invasive prognostic damage prediction model based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels includes: training a preset predictor using the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels to construct the corresponding non-invasive prognostic damage prediction model. The preset predictor can be configured as a deep learning-based prediction network (e.g., CNN, RNN, GCN, etc.) or a machine learning-based predictor (e.g., SVM, Random Forest, XGBoost, etc.).

[0028] In this embodiment of the disclosure, determining the preset peripheral blood index includes: acquiring multiple peripheral blood indices to be processed corresponding to the cross-sectional pathological prediction cohort; performing logistic regression analysis on the multiple peripheral blood indices to be processed and their corresponding IgA nephropathy prognostic pathological damage labels to obtain multiple levels of significance; configuring the peripheral blood indices to be processed corresponding to the multiple levels of significance less than the preset level of significance as preset peripheral blood indices; and / or, configuring the prognostic damage and the IgA nephropathy prognostic pathological damage labels to reflect renal tubular atrophy and interstitial fibrosis corresponding to chronic irreversible damage or glomerular endothelial cell hyperplasia corresponding to acute active inflammation.

[0029] Specifically, the multiple peripheral blood indicators to be processed are configured as one or more of the following indicators: red blood cell distribution width, platelet index, neutrophil index, lymphocyte index, hemoglobin index, low-density lipoprotein index, total cholesterol (blood lipid) index, triglyceride index, blood glucose index, complement C4 index, IgM index, IgA index, IgG index, complement C3 index, D-dimer (coagulation) index, CRP index, erythrocyte sedimentation rate (ESR) index, albumin index, photostatin C index, uric acid index, creatinine index, blood urea nitrogen index, eGFR index, and presence of urinary red blood cells.

[0030] Specifically, it also includes: acquiring multiple unprocessed urine indicators corresponding to the cross-sectional pathological prediction queue; performing logistic regression analysis on the multiple unprocessed urine indicators and their corresponding IgA nephropathy prognostic pathological damage labels to obtain multiple corresponding significance levels; and configuring the unprocessed urine indicators with multiple significance levels less than a preset significance level as preset urine indicators.

[0031] Specifically, based on the preset peripheral blood indicators, preset urine indicators, and the IgA nephropathy prognostic pathological damage labels, a corresponding non-invasive predictive model for prognostic damage is constructed. Based on this non-invasive predictive model, the preset peripheral blood indicators and preset urine indicators to be predicted are used to predict prognostic damage. The multiple urine indicators to be processed can be configured as 24-hour urine protein quantification, urine alpha-microglobulin, urine β2-microglobulin, and urine ACR.

[0032] Specifically, some malignant tumors (such as renal cell carcinoma, lung cancer, lymphoma, etc.) may cause "tumor-associated glomerular disease" by producing abnormal antibodies or immune complexes, which may then manifest as IgA nephropathy.

[0033] In this embodiment, the IgA nephropathy is configured as IgA nephropathy induced by a tumor lesion. The prediction of tumor-induced IgA nephropathy includes: training a preset predictor using quantitative imaging features of a preset tumor lesion corresponding to a tumor lesion in a medical image of a preset location at a first time point, and pathological results corresponding to induced or non-induced IgA nephropathy at a second time point after the first time point, to construct a tumor-induced IgA nephropathy prediction model; and using the tumor-induced IgA nephropathy prediction model and the quantitative imaging features of the preset tumor lesion corresponding to a tumor lesion in a medical image of a preset location to be predicted, performing a prediction of tumor-induced IgA nephropathy. This addresses the technical problem of the lack of accurate, effective, and non-invasive preoperative assessment methods for tumor-induced IgA nephropathy.

[0034] Specifically, the preset predictor can be configured as a prediction network based on deep learning (e.g., CNN, RNN, GCN, etc.) or a predictor based on machine learning (e.g., SVM, Random Forest, XGBoost, etc.).

[0035] In this embodiment of the disclosure, before training the preset predictor, the method includes: extracting quantitative image features of tumor lesions corresponding to tumor lesions in medical images of preset locations; and selecting preset quantitative image features of tumor lesions related to the pathological results from the quantitative image features of the tumor lesions to be selected.

[0036] In this embodiment of the disclosure, determining the preset quantitative imaging features of tumor lesions related to the pathological results includes: using a univariate logistic regression analysis algorithm, the quantitative imaging features of the tumor lesions to be selected and their corresponding pathological results, performing univariate logistic regression on each of the quantitative imaging features of the tumor lesions to be selected, and determining the preset quantitative imaging features of tumor lesions related to the pathological results.

[0037] In this embodiment of the disclosure, the quantitative imaging features of the tumor lesion to be selected include one or more of the following: the long diameter of the tumor lesion, the long diameter of the solid component of the tumor lesion, the ratio of the solid component of the tumor lesion to the total tumor lesion, the location of the tumor lesion, whether the edge of the tumor lesion is clear, the morphology of the tumor lesion, the spinous process sign of the tumor lesion, the lobulation sign of the tumor lesion, the pleural traction sign of the tumor lesion, the cavitation sign of the tumor lesion, and the air bronchus sign of the tumor lesion.

[0038] In this embodiment of the disclosure, determining whether the edge of the tumor lesion corresponding to the quantitative imaging features of the selected tumor lesion is clear includes: performing an erosion operation on the tumor lesion mask image to obtain a tumor lesion mask eroded image; subtracting the tumor lesion mask eroded image from the tumor lesion mask image to obtain a tumor lesion mask processed image; mapping the tumor lesion location coordinates in the tumor lesion mask processed image to the preset site medical image; calculating multiple gradients corresponding to the tumor lesion location coordinates in the preset site medical image; if the gradient magnitude values ​​corresponding to the multiple tumor lesions are greater than or equal to a preset gradient magnitude value, then the boundary of the tumor lesion location corresponding to the tumor lesion location greater than or equal to the preset gradient magnitude value is determined to be clear; otherwise, the tumor lesion boundary is determined to be unclear.

[0039] In this embodiment of the disclosure, determining the spur includes: detecting the edges of the tumor lesion mask in the tumor lesion mask image to obtain the tumor lesion mask edge line; calculating the first pixel distance and the second pixel distance between any tumor lesion mask edge pixel on the tumor lesion mask edge line and its adjacent tumor lesion mask edge pixels; if the first pixel distance and the second pixel distance are greater than a set pixel distance, then the arbitrary tumor lesion mask edge pixel is configured as a spur; and / or, determining the lobulation sign of the tumor lesion includes: performing an erosion operation on the tumor lesion mask image corresponding to each tumor lesion in the preset site medical image to obtain a tumor lesion mask erosion image; subtracting the corresponding tumor lesion mask erosion image from the tumor lesion mask image of each tumor lesion to obtain a tumor lesion mask processing image corresponding to each tumor lesion; judging the morphology of the tumor lesion in the tumor lesion mask processing image corresponding to each tumor lesion; if the morphology of the tumor lesion is irregular, then determining whether there is a tumor lesion lobulation sign in the tumor lesion mask processing image corresponding to each tumor lesion.

[0040] Specifically, before training the preset predictor, the process includes: extracting quantitative imaging features of tumor lesions corresponding to tumor lesions in medical images of preset locations; and selecting preset quantitative imaging features of tumor lesions that are related to the pathological results from the quantitative imaging features of the tumor lesions to be selected.

[0041] Specifically, determining the preset quantitative imaging features of tumor lesions related to the pathological results includes: using a univariate logistic regression analysis algorithm, the quantitative imaging features of the selected tumor lesions and their corresponding pathological results, performing univariate logistic regression on each of the quantitative imaging features of the selected tumor lesions to determine the preset quantitative imaging features of tumor lesions related to the pathological results.

[0042] Specifically, the quantitative imaging features of the tumor lesions to be selected include one or more of the following: the long diameter of the tumor lesion, the long diameter of the solid component of the tumor lesion, the ratio of the solid component of the tumor lesion to the total tumor lesion, the location of the tumor lesion, whether the edge of the tumor lesion is clear, the morphology of the tumor lesion, the spinous process sign of the tumor lesion, the lobulation sign of the tumor lesion, the pleural traction sign of the tumor lesion, the cavitation sign of the tumor lesion, and the air bronchus sign of the tumor lesion.

[0043] Specifically, the step of extracting quantitative imaging features of tumor lesions corresponding to tumor lesions in medical images of preset locations includes: segmenting the tumor lesions corresponding to tumor lesions in medical images of preset locations to obtain a tumor lesion mask image; and based on the tumor lesion mask image, determining one or more of the following: the long diameter of the tumor lesion, whether the edge of the tumor lesion is clear, the morphology of the tumor lesion, the spinous process sign of the tumor lesion, the lobulation sign of the tumor lesion, and the vacuolation sign of the tumor lesion corresponding to the quantitative imaging features of the tumor lesion to be selected.

[0044] Specifically, before training the preset predictor, the numerical quantitative image features of the selected tumor lesions are standardized or normalized, and the categorical quantitative image features of the selected tumor lesions are encoded.

[0045] Specifically, for example, before training the preset predictor, the quantitative image features of the tumor lesion corresponding to the long diameter of the tumor lesion, the long diameter of the solid component of the tumor lesion, and the proportion of the solid component of the tumor lesion to the tumor lesion are standardized or normalized to eliminate the influence of different dimensions on the training of the preset predictor.

[0046] Specifically, based on the tumor lesion mask image, after determining one or more of the following quantitative imaging features corresponding to the tumor lesion to be selected: tumor location, clarity of tumor edge, morphology of tumor lesion, presence of spinous process sign, presence of lobulation sign, presence of pleural traction sign, presence of cavitation sign, and presence of air bronchus sign, the tumor lesion to be selected is then encoded to obtain the tumor lesion to be selected. The quantitative imaging features of the lesions correspond to one or more of the following coding features: tumor lesion location, clarity of tumor lesion margins, tumor lesion morphology, presence of spinous process sign, presence of lobulation sign, presence of pleural traction sign, presence of cavitation sign, and presence of air bronchus sign. These coding features are then used to train a preset predictor, thereby constructing a tumor lesion-induced IgA nephropathy prediction model.

[0047] Specifically, the locations of tumor lesions corresponding to the quantitative imaging features of the selected tumor lesions are encoded to obtain different tumor lesion location encoding features corresponding to a first numerical queue (0, 1, 2, ... or 0, 2, 4, ...) for different tumor lesion locations; wherein, the different tumor lesion location encoding features corresponding to the first numerical queue (0, 1, 2, ... or 0, 2, 4, ...) for different tumor lesion locations are configured according to a predefined lung field anatomical region. Similarly, the morphology of tumor lesions corresponding to the quantitative imaging features of the selected tumor lesions is encoded to obtain different tumor lesion morphology encoding features corresponding to a second numerical queue (0, 1, 2, ... or 1, 3, 5, ...); wherein, the second numerical queue (0, 1, 2, ... or 1, 3, 5, ...) corresponding to the different tumor lesion morphology encoding features are configured according to different types corresponding to the preset closed curve equation or the preset closed surface equation.

[0048] Specifically, the quantitative imaging features of the selected tumor lesions are encoded based on one or more of the following: whether the tumor lesion margin is clear, whether the tumor lesion has a spiky sign, whether the tumor lesion has a lobulated sign, whether the tumor lesion has a pleural traction sign, whether the tumor lesion has a vacuolar sign, and whether the tumor lesion has an air bronchus sign. This includes: if the tumor lesion is present, the corresponding quantitative imaging feature is configured with a first value (e.g., 1); otherwise, the corresponding quantitative imaging feature is configured with a second value different from the first value (e.g., 0). For example, if the tumor lesion margin is clear, then when encoding the clear tumor lesion margin, the clear tumor lesion margin is configured with the first value; if the tumor lesion has a spiky sign, then when encoding the spiky sign, the spiky sign is configured with the first value.

[0049] Specifically, based on the tumor lesion mask image, determining whether the edge of the tumor lesion corresponding to the quantitative imaging features of the selected tumor lesion is clear includes: performing an erosion operation on the tumor lesion mask image to obtain a tumor lesion mask erosion image; subtracting the tumor lesion mask erosion image from the tumor lesion mask image to obtain a tumor lesion mask processed image; mapping the tumor lesion location coordinates in the tumor lesion mask processed image to the preset site medical image; calculating multiple gradients corresponding to the tumor lesion location coordinates in the preset site medical image; if the gradient magnitude values ​​corresponding to the multiple tumor lesions are greater than or equal to a preset gradient magnitude value, then the boundary of the tumor lesion location corresponding to the tumor lesion location greater than or equal to the preset gradient magnitude value is determined to be clear; otherwise, the tumor lesion boundary is determined to be unclear.

[0050] Specifically, the step of calculating multiple gradients corresponding to the location coordinates of the tumor lesion in the medical image of the preset location includes: determining multiple first position coordinates and multiple second position coordinates on both sides of the location coordinates of the tumor lesion in the medical image of the preset location, corresponding to the x-direction coordinates and the y-direction coordinates respectively; calculating the difference between the first pixel value corresponding to the multiple first position coordinates and the second pixel value corresponding to the multiple second position coordinates respectively to obtain multiple x-direction gradients corresponding to the location coordinates of the tumor lesion; calculating the difference between the third pixel value corresponding to the multiple third position coordinates and the fourth pixel value corresponding to the multiple fourth position coordinates respectively to obtain multiple y-direction gradients corresponding to the location coordinates of the tumor lesion; and configuring the multiple x-direction gradients and their corresponding multiple y-direction gradients as multiple gradients corresponding to the location coordinates of the tumor lesion.

[0051] Specifically, based on the location coordinates of the tumor lesion, determining multiple first location coordinates and multiple second location coordinates on both sides of the tumor lesion location coordinates in the preset location medical image in the x direction includes: using the tumor lesion location coordinates, determining multiple first location coordinates and multiple second location coordinates on both sides in the x direction according to a preset distance.

[0052] Specifically, based on the location coordinates of the tumor lesion, determining multiple third position coordinates and multiple fourth position coordinates on both sides of the tumor lesion location coordinates in the preset location medical image in the y direction includes: using the tumor lesion location coordinates, determining multiple third position coordinates and multiple fourth position coordinates on both sides in the x direction according to a preset distance.

[0053] Specifically, based on the tumor lesion mask image, determining the morphology of the tumor lesion corresponding to the quantitative imaging features of the selected tumor lesion includes: determining the tumor lesion mask processing image corresponding to each tumor lesion in the tumor lesion mask image of the medical image of the preset location; extracting multiple tumor lesion mask position coordinates corresponding to the tumor lesion mask line in the tumor lesion mask processing image corresponding to each tumor lesion; fitting or reconstructing the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion corresponding to each tumor lesion.

[0054] Specifically, the step of fitting or reconstructing the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion includes: under the optimal criterion, approximating multiple preset closed curve equations or multiple preset closed surface equations using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion corresponding to each tumor lesion.

[0055] Specifically, under the optimal criterion, the step of approximating a preset closed curve equation or a preset closed surface equation using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion corresponding to each tumor lesion includes: under the optimal criterion, minimizing the error or absolute residual between the approximation equation and the multiple preset closed curve equation or multiple preset closed surface equation when the multiple tumor lesion mask position coordinates corresponding to each tumor lesion approximate the multiple preset closed curve equation or multiple preset closed surface equation, thereby determining the morphology of the tumor lesion corresponding to each tumor lesion.

[0056] Specifically, under the optimal criterion, the step of approximating a preset closed curve equation or a preset closed surface equation using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of each tumor lesion includes: if the preset site medical image is configured as a two-dimensional medical image, then under the optimal criterion, approximating multiple preset closed curve equations using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of each tumor lesion; if the preset site medical image is configured as a three-dimensional medical image, then under the optimal criterion, approximating multiple preset closed surface equations using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of each tumor lesion. The optimal criterion is configured as one or more of the least squares criterion, the minimum absolute residual criterion, etc.

[0057] Specifically, the optimal criterion is configured as the least squares criterion. Then, after minimizing the errors between the approximation equations and the approximation equations of multiple tumor lesion mask positions when approximating multiple preset closed curve equations or multiple preset closed surface equations for each tumor lesion, the morphologies of multiple tumor lesions corresponding to each tumor lesion are obtained. The errors in the tumor lesion mask positions corresponding to each of the multiple tumor lesion morphologies are extracted. The tumor lesion morphology with the smallest error among the multiple tumor lesion mask positions corresponding to the multiple tumor lesion morphologies is configured as the final tumor lesion morphology corresponding to each tumor lesion.

[0058] Specifically, the optimal criterion is configured as the least squares criterion. Then, after minimizing the absolute residuals of the approximation equations (which approximate multiple preset closed curve equations or multiple preset closed surface equations) and the tumor lesion mask position coordinates (which approximate multiple preset closed curve equations or multiple preset closed surface equations) for each tumor lesion, the morphologies of multiple tumor lesions corresponding to each tumor lesion are obtained. The absolute residuals of the tumor lesion mask position coordinates corresponding to each of the multiple tumor lesion morphologies are extracted. The tumor lesion morphology corresponding to the morphology with the smallest error in the absolute residuals of the tumor lesion mask position coordinates corresponding to the multiple tumor lesion morphologies is configured as the final tumor lesion morphology corresponding to each tumor lesion.

[0059] Specifically, the preset closed curve equation is configured as one or more of the two-dimensional closed curves corresponding to first-order polynomials, second-order polynomials, or higher-order polynomials, which are the basis of polynomial fitting. For example, the preset closed curve equation is configured as a circle, ellipse, or conic section corresponding to a basic geometric closed curve; a triangle, quadrilateral, pentagon, or hexagon corresponding to a closed curve enclosed by line segments; a spline curve; a polyline closed curve, etc. The preset closed surface equation is configured as a bivariate polynomial or a multivariate polynomial. For example, the preset closed surface equation is configured as a circular surface, elliptical surface, or conic surface corresponding to a basic geometric closed surface; a triangular surface, quadrilateral surface, pentagonal surface, or hexagonal surface corresponding to a closed surface enclosed by line segments; a closed surface composed of multiple surfaces, etc.

[0060] Specifically, determining the tumor lesion mask image includes: performing an erosion operation on the tumor lesion mask image to obtain an eroded tumor lesion mask image; and subtracting the eroded tumor lesion mask image from the tumor lesion mask image to obtain a processed tumor lesion mask image. Prior to determining the processed tumor lesion mask image corresponding to the tumor lesion in the tumor lesion mask image of the preset medical image, the process includes: using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network to segment the tumor lesions in the preset medical image, thereby determining the tumor lesion mask image of the preset medical image.

[0061] Specifically, before segmenting tumor lesions in medical images of preset locations using a preset segmentation model constructed based on a deep learning segmentation network, the process includes: training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain a preset tumor lesion segmentation model constructed based on the deep learning segmentation network. Specifically, before training the preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain the preset tumor lesion segmentation model constructed based on the deep learning segmentation network, the process includes: using segmentation annotation software to delineate tumor lesions in a preset number of medical images of preset locations to obtain corresponding tumor lesion mask training images.

[0062] Specifically, before training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain a preset segmentation model constructed based on the deep learning segmentation network, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions on the preset number of medical images of preset locations, obtaining corresponding tumor lesion mask training images. Those skilled in the art can configure the preset number according to actual needs. For example, the preset number can be configured as 300-500.

[0063] Specifically, the preset site medical image can be configured as one or more of the following medical images: chest X-ray image / chest CT image / chest PET image / chest MRI image; knee joint X-ray image / knee joint CT image / knee joint PET image / knee joint MRI image; spine X-ray image / spine CT image / spine PET image / spine MRI image; breast X-ray image / breast CT image / breast PET image / breast MRI image; abdomen X-ray image / abdomen CT image / abdomen PET image / abdomen MRI image; vertebral body X-ray / vertebral body CT abdominal / vertebral body PET abdominal / vertebral body MRI image; breast ultrasound image; and abdominal ultrasound image. The lesion corresponding to the preset site medical image is configured as one or more of the following lesions on the medical image: pulmonary nodule, lung cancer, liver cancer, breast cancer, lipoma, fibroma, hamartoma, etc.

[0064] Specifically, the lesions corresponding to the preset medical images, such as chest X-ray images, chest CT images, chest PET images, and chest MRI images, are configured as one or more of the lesions such as pulmonary nodules and lung cancer on the medical images, which will be further described in detail.

[0065] Specifically, determining the long diameter of the tumor lesion corresponding to the quantitative imaging features of the selected tumor lesion includes: performing edge detection on each tumor lesion in the mask image of the tumor lesion corresponding to the medical image of the preset location to obtain the mask edge line of each tumor lesion; setting a reference mask point on the mask edge line of each tumor lesion respectively; calculating multiple distances from the reference mask point to multiple mask points on the mask edge line of each tumor lesion excluding the reference mask point; and configuring the maximum distance among the multiple distances as the long diameter of each tumor lesion.

[0066] Specifically, determining the pleural traction sign of the tumor lesion includes: detecting the pleural region of the tumor lesion at a predetermined location in a lung field within a medical image of a predetermined site, and the pleura adjacent to the tumor lesion at that predetermined location; and determining, based on the pleural region image of the tumor lesion, whether the pleural traction sign of the tumor lesion exists in the medical image of the predetermined site. This addresses at least one of the technical problems of existing methods that require manual determination of the pleural traction sign of tumor lesions, such as excessive subjectivity, lack of intelligent determination of the pleural traction sign, resulting in a huge workload for doctors, and preventing further technical adjustments such as surgery and prediction of IgA nephropathy induced by tumor lesions.

[0067] Specifically, determining the tumor lesion at the preset location in the lung field includes: segmenting the tumor lesion in the medical image of the preset location using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network to obtain a tumor lesion mask image corresponding to the medical image of the preset location; and determining the tumor lesion at the preset location in the lung field based on the tumor lesion location coordinates in the tumor lesion mask image.

[0068] Specifically, determining the tumor lesion at a preset location in the lung field based on the tumor lesion location coordinates in the tumor lesion mask image includes: using a lung field dynamic region segmentation method to segment the medical image of the preset location into lung field regions, obtaining a lung field region segmentation network corresponding to each tumor lesion; locating the lung field anatomical region corresponding to each tumor lesion based on its location and the corresponding lung field region segmentation network; determining the tumor lesion at a preset location in the lung field based on the lung field anatomical region corresponding to each tumor lesion; and / or, determining the tumor lesion at a preset location in the lung field based on the lung field anatomical region corresponding to each tumor lesion includes: if the lung field anatomical region corresponding to the tumor lesion is close to the lung field boundary, then the tumor lesion close to the lung field boundary is determined as the tumor lesion at the preset location in the lung field.

[0069] Specifically, determining whether the lung field anatomical region corresponding to the tumor lesion is close to the lung field boundary includes: calculating multiple distances between each tumor lesion and the lung field boundary in the lung field mask image corresponding to the preset site medical image; and determining whether the lung field anatomical region corresponding to the tumor lesion is close to the lung field boundary based on the multiple distances between the lung field boundaries and the preset distance.

[0070] Specifically, determining whether the anatomical region of the lung field corresponding to the tumor lesion is close to the lung field boundary based on the distances between the lung field boundaries of the multiple tumor lesions and a preset distance includes: if the distances between the lung field boundaries of the multiple tumor lesions are less than or equal to the preset distance, then the anatomical region of the lung field corresponding to the tumor lesion is determined to be close to the lung field boundary; otherwise, the anatomical region of the lung field corresponding to the tumor lesion is determined to be far from the lung field boundary. Those skilled in the art can determine the preset distance according to actual needs. For example, the preset distance can be configured as 1-10 pixels or other pixel values.

[0071] Specifically, before segmenting the medical images of the preset locations using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network, the process includes: training the preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain the preset tumor lesion segmentation model constructed based on the deep learning segmentation network. Specifically, before training the preset deep learning segmentation network, the process includes: using segmentation annotation software to delineate tumor lesions on a preset number of medical images of preset locations to obtain corresponding tumor lesion mask training images.

[0072] Specifically, before training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain a preset segmentation model constructed based on the deep learning segmentation network, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions on the preset number of medical images of preset locations, obtaining corresponding tumor lesion mask training images. Those skilled in the art can configure the preset number according to actual needs. For example, the preset number can be configured as 300-500.

[0073] Specifically, determining the tumor lesion at a predetermined location in the lung field includes: segmenting the lung field in the medical image of the predetermined location using a predetermined lung field segmentation model constructed based on a deep learning segmentation network to obtain a lung field mask image corresponding to the medical image of the predetermined location; extracting the lung field of the lung field mask image by a first predetermined edge width to obtain a lung field image with a first predetermined edge width; segmenting the tumor lesion in the lung field image with the first predetermined edge width using a predetermined tumor lesion segmentation model constructed based on a deep learning segmentation network to obtain a tumor lesion mask image with a first predetermined edge width; and determining the tumor lesion in the tumor lesion mask image with the first predetermined edge width as the tumor lesion at the predetermined location in the lung field. This addresses at least one of the technical problems of excessive subjectivity in tumor lesion location positioning, the huge workload for doctors caused by the lack of intelligent tumor lesion location positioning, and the inability to further predict IgA nephropathy induced by tumor lesions.

[0074] Specifically, the step of extracting the lung field from the lung field mask image with a first preset edge width to obtain a lung field image with a first preset edge width includes: performing an erosion operation on the lung field in the lung field mask image using the first preset edge width to obtain a lung field erosion mask image; performing a pixel-level subtraction operation on the lung field mask image and the lung field erosion mask image to obtain a lung field mask image with a first preset edge width; and performing a pixel-level multiplication operation on the lung field mask image with the preset medical image to obtain a lung field image with a first preset edge width. More specifically, the mask value corresponding to the lung field in the lung field mask image can be configured to 1. Those skilled in the art can determine the first preset edge width according to actual needs. For example, the first preset edge width can be configured to 1-10 pixels or other pixel values.

[0075] Specifically, before segmenting the medical images of the preset locations using a preset lung field segmentation model constructed based on a deep learning segmentation network, the process includes: training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding lung field mask training images to obtain a preset lung field segmentation model constructed based on the deep learning segmentation network.

[0076] Specifically, before training the preset deep learning segmentation network, the process includes: using segmentation annotation software to delineate the lung fields of a preset number of preset medical images to obtain corresponding lung field mask training images.

[0077] Specifically, before training a preset deep learning segmentation network using a preset number of preset site medical images and their corresponding lung field mask training images to obtain a preset segmentation model constructed based on the deep learning segmentation network, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate the lung fields of the preset number of preset site medical images to obtain corresponding lung field mask training images. Those skilled in the art can configure the preset number according to actual needs.

[0078] Specifically, determining the pleural mask image corresponding to the pleura includes: expanding the lung field in the lung field mask image using a second preset edge width to obtain an expanded lung field mask image; performing pixel-level subtraction on the expanded lung field mask image and the lung field mask image to determine the pleural mask image corresponding to the pleura; this addresses the technical problem that the pleura is not clearly displayed on the medical image of the preset location, making it difficult to delineate the pleura, thus preventing the acquisition of a preset pleural segmentation model based on a deep learning segmentation network that meets performance requirements.

[0079] Specifically, the detection of the pleural region image of the tumor lesion at a predetermined location in the lung field of the medical image of the predetermined location includes: drawing multiple rays corresponding to multiple predetermined directions from the boundary line or surface of the tumor lesion at the predetermined location to the boundary line or surface of the pleural mask adjacent to the pleural lesion at the predetermined location; and defining the image area enclosed by the rays intersecting the pleural mask boundary line or surface of the predetermined location, the boundary line or surface of the tumor lesion at the predetermined location, and the boundary line or surface of the pleural mask adjacent to the pleural lesion at the predetermined location as the pleural region image of the tumor lesion.

[0080] Specifically, determining the image area enclosed by the rays intersecting the multiple rays with the pleural mask boundary line or pleural mask surface, the tumor lesion boundary line or tumor lesion surface at the preset location, and the pleural mask boundary line or pleural mask surface adjacent to the tumor lesion at the preset location as the tumor lesion pleural region image includes: determining the image area enclosed by the rays intersecting the multiple rays with the pleural mask boundary line or pleural mask surface, the tumor lesion boundary line or tumor lesion surface at the preset location, and the pleural mask boundary line or pleural mask surface adjacent to the tumor lesion at the preset location as the tumor lesion pleural region mask image; and performing a pixel-level multiplication operation between the tumor lesion pleural region mask image and the medical image of the preset location to determine the tumor lesion pleural region image.

[0081] Specifically, if the preset site medical image is configured as a two-dimensional medical image, then detecting the pleural region image of the tumor lesion between the preset location tumor lesion in the lung field and the adjacent pleura of the preset location tumor lesion in the preset site medical image includes: drawing multiple rays corresponding to multiple predetermined directions from the tumor lesion boundary line of the preset location tumor lesion to the pleural mask boundary line of the adjacent pleura of the preset location tumor lesion; and determining the area enclosed by the rays intersecting the pleural mask boundary line, the tumor lesion boundary line of the preset location tumor lesion, and the pleural mask boundary line of the adjacent pleura of the preset location tumor lesion as the pleural region image of the tumor lesion.

[0082] Specifically, the step of drawing multiple rays corresponding to multiple predetermined directions from the tumor lesion boundary line at the preset location to the pleural mask boundary line adjacent to the tumor lesion at the preset location includes: setting an initial radiation point at the tumor lesion boundary line at the preset location; determining multiple radiation points on the tumor lesion boundary line based on the initial radiation point and the preset radiation point step distance; and drawing multiple rays corresponding to multiple predetermined two-dimensional directions starting from the multiple radiation points.

[0083] Specifically, if the preset site medical image is configured as a three-dimensional medical image, then detecting the pleural region image of the tumor lesion between the preset location tumor lesion in the lung field and the adjacent pleura of the preset location tumor lesion in the preset site medical image includes: drawing multiple rays corresponding to multiple predetermined directions from the surface of the preset location tumor lesion to the pleural mask surface adjacent to the preset location tumor lesion; determining the area enclosed by the rays intersecting the multiple rays and the pleural mask surface, the surface of the preset location tumor lesion, and the adjacent pleura or the pleural mask surface of the preset location tumor lesion as the pleural region image of the tumor lesion; and / or, The step of casting multiple rays in multiple predetermined directions from the surface of the tumor lesion at the predetermined location to a pleural mask surface adjacent to the pleura at the predetermined location includes: determining a side of the pleural mask surface corresponding to the lung field anatomical region adjacent to the tumor lesion at the predetermined location; setting an initial radiation point on the side of the pleural mask surface; determining multiple radiation points on the side of the pleural mask surface based on the initial radiation point and the predetermined radiation point step distance; and casting multiple rays in multiple predetermined three-dimensional directions starting from the multiple radiation points.

[0084] Specifically, based on the pleural mask image corresponding to the pleura, the corresponding pleural mask boundary line or pleural mask surface is determined. Specifically, determining the corresponding pleural mask boundary line or pleural mask surface based on the pleural mask image corresponding to the pleura includes: if the preset medical image is configured as a two-dimensional medical image, then edge detection is performed on the pleural mask image corresponding to the pleura to determine the corresponding pleural mask boundary line; if the preset medical image is configured as a three-dimensional medical image, then edge detection is performed on each layer of the pleural mask image corresponding to the pleura to obtain an edge image of each layer of the pleural mask; and three-dimensional reconstruction is performed on each layer of the pleural mask edge image to determine the corresponding pleural mask surface.

[0085] Specifically, determining the pleural mask surface corresponding to the lung field anatomical region adjacent to the preset location tumor lesion includes: dividing the pleura into left lung pleura and right lung pleura according to the lung field anatomical region of the preset location tumor lesion; wherein the left lung pleura and right lung pleura correspond to the left lung pleural mask surface and the right lung pleural mask surface, respectively. Therefore, before determining the pleural mask surface corresponding to the lung field anatomical region adjacent to the preset location tumor lesion, the process includes: dividing the pleura into multiple lung field anatomical pleural mask surfaces according to the lung field anatomical region of the preset location tumor lesion; determining the pleural mask surface corresponding to the lung field anatomical region adjacent to the preset location tumor lesion according to the multiple lung field anatomical pleural mask surfaces; setting initial radiation points on the pleural mask surface; and determining multiple radiation points on the pleural mask surface using the initial radiation points and the preset radiation point step distance.

[0086] For example, based on the lung field anatomical region of the tumor lesion at the preset location, the pleura is divided into the left lung pleura and the right lung pleura; a left lung pleura mask surface or a right lung pleura mask surface corresponding to the lung field anatomical region adjacent to the tumor lesion at the preset location is determined; an initial radiation point is set on the left lung pleura mask surface or the right lung pleura mask surface; multiple radiation points on the pleura mask surface are determined based on the initial radiation point and the preset radiation point step distance.

[0087] Specifically, if the lung field anatomical region of the tumor lesion at the preset location is close to the anterior side of the left lung pleura, then the left lung pleural mask surface corresponding to the anterior side of the left lung pleura is retained; if the lung field anatomical region of the tumor lesion at the preset location is close to the anterior side of the right lung pleura, then the right lung pleural mask surface corresponding to the anterior side of the left lung pleura is retained. Based on the above, those skilled in the art can determine the specific pleural mask surface according to the actual anatomical structure of the lung field. Therefore, the pleural mask surface can also be configured as a pleural mask surface corresponding to different lung lobes. For example, the pleural mask surface of any side of the right lung lobe.

[0088] Specifically, those skilled in the art can determine the second preset edge width according to actual needs. For example, the second preset edge width can be configured to 1-5 pixels or other pixel values.

[0089] Specifically, determining whether there is a pleural traction sign of the tumor lesion in the medical image of the preset location based on the pleural region image of the tumor lesion includes: detecting whether there are cord-like or linear features between the adjacent pleura of the tumor lesion at the preset location in the pleural region image of the tumor lesion; if so, determining that there is a pleural traction sign of the tumor lesion in the medical image of the preset location; otherwise, determining that there is no pleural traction sign of the tumor lesion in the medical image of the preset location.

[0090] Specifically, detecting whether there are cord-like or linear features between the pleura adjacent to the tumor lesion at the preset location in the image of the tumor lesion pleural region includes: using a preset feature target detection model based on deep learning to perform feature target detection on the image of the tumor lesion pleural region, and determining whether there are cord-like or linear features between the pleura adjacent to the tumor lesion at the preset location in the image of the tumor lesion pleural region.

[0091] Specifically, before performing feature target detection on the pleural region image of the tumor lesion using a preset feature target detection model based on deep learning, the process includes: training feature point training images using a preset number of preset location medical training images and their corresponding strip-like features and / or line-like features, training a preset deep learning target detection network, and obtaining a preset feature target detection model constructed based on the deep learning target detection network.

[0092] Specifically, before training the preset deep learning object detection network, the process includes: using segmentation and annotation software to delineate the corresponding linear and / or bar-shaped feature bounding boxes on a preset number of medical images of preset body parts, thereby obtaining corresponding linear and / or bar-shaped feature training images. For example, annotation software such as CasiaLabeler or Labelme can be used to delineate the corresponding linear and / or bar-shaped feature bounding boxes on a preset number of medical training images of preset body parts, thereby obtaining corresponding linear and / or bar-shaped feature training images. Those skilled in the art can configure the preset number according to actual needs. For example, the preset number can be configured as 300-500 or other possible values. The preset deep learning object detection network can be configured as a YOLO series or an object detection network based on YOLO.

[0093] Specifically, determining the proportion of solid tumor lesions includes: segmenting the medical image of the preset location using a deep learning-based solid tumor lesion segmentation model to obtain a tumor lesion segmentation mask image and a solid tumor lesion segmentation mask image; and determining the proportion of solid tumor lesions within the tumor lesion based on the tumor lesion segmentation mask image and the solid tumor lesion segmentation mask image.

[0094] Specifically, the step of segmenting the medical image of the preset location using a deep learning-based solid tumor lesion segmentation model to obtain a tumor lesion segmentation mask image and a solid tumor lesion segmentation mask image includes: constructing a training image of the tumor lesion in each set of medical training images of the preset location and a binary mixed location mask corresponding to the solid tumor lesion within the tumor lesion; training the preset deep learning segmentation network using the binary mixed location mask training image, the solid tumor lesion edge mask training image corresponding to the solid tumor lesion, and the corresponding preset location medical training image to obtain a solid tumor lesion segmentation model; and segmenting the medical image of the preset location to be segmented based on the solid tumor lesion segmentation model to obtain a tumor lesion segmentation mask image and a solid tumor lesion segmentation mask image. This addresses at least one of the technical problems of the current method of manually segmenting solid tumor lesions, which results in a huge workload for doctors and prevents further technical work such as surgery and prediction of IgA nephropathy induced by tumor lesions.

[0095] Specifically, unlike existing technologies, a set of medical training images for each preset location is constructed, containing tumor lesions and corresponding binary mixed location mask training images of solid tumor lesions within the preset location. During the training of the preset deep learning segmentation network, the network pays more attention to the anatomical relationship between the tumor lesions and the solid tumor lesions within the preset location. This leads to collaborative learning and response between the tumor lesions and the solid tumor lesions within the preset location, thereby enabling effective anatomical segmentation of the solid tumor lesions within the preset location and creating anatomical constraints on the segmentation of the solid tumor lesions within the preset location.

[0096] Specifically, when each set of preset site medical training images and the preset site medical images to be segmented in the preset site medical training image set can be configured as two-dimensional preset site medical images, each set of preset site medical training images may be one preset site medical training image corresponding to a subject at a specific moment, or multiple preset site medical training images corresponding to subjects at multiple specific moments. For example, when the two-dimensional preset site medical images are configured as preset site X-ray images or preset site ultrasound images, each set of preset site medical training images may be one preset site medical training image corresponding to a subject at a specific moment corresponding to breath-holding, deep breathing, or deep inspiration, or multiple preset site medical training images corresponding to subjects at multiple specific moments corresponding to the breathing process or breath-holding process.

[0097] Specifically, when each set of preset site medical training images and the preset site medical images to be segmented in the preset site medical training image set can be configured as three-dimensional preset site medical images, each set of preset site medical training images may be multiple three-dimensional preset site medical training images corresponding to the examinee at a specific time, or multiple four-dimensional preset site medical training images corresponding to the examinee at multiple specific times. For example, when the three-dimensional preset site medical images can be configured as one or more of preset site CT images, preset site MRI images, preset site PET images, preset site CT-PET images, and preset site MRI-PET images, each set of preset site medical training images may be multiple three-dimensional preset site medical training images corresponding to the examinee at a specific time corresponding to breath-holding, deep breathing, or deep inspiration, or multiple four-dimensional preset site medical training images corresponding to the examinee at multiple specific times corresponding to the breathing process or breath-holding process.

[0098] Specifically, the preset deep learning segmentation network is configured as one or more deep learning segmentation networks such as Fully Convolutional Network (FCN), U-Net, DeepLab, and Pyramid Spatial Pooling Network (PSPNet).

[0099] Specifically, each set of preset site medical training images may include multiple tumor lesions, and each tumor lesion may include solid tumor lesions. In other words, the multiple tumor lesions in each set of preset site medical training images may or may not contain solid tumor lesions.

[0100] Specifically, the construction of a binary mixed location mask training image for each set of preset site medical training images in the preset site medical training image set, including: extracting a tumor lesion mask training image with a first mask value corresponding to the tumor lesion in each set of preset site medical training images in the preset site medical training image set, and a solid tumor lesion mask training image with a second mask value corresponding to the solid tumor lesion in the solid tumor lesion; mapping the second mask value to the tumor lesion position in the solid tumor lesion mask training image according to the position of the solid tumor lesion in the solid tumor lesion mask training image, so as to construct a binary mixed location mask training image.

[0101] Specifically, determining the solid tumor lesion edge mask training image corresponding to the solid tumor lesion includes: performing an erosion operation on the solid tumor lesion mask training image corresponding to each set of preset site medical training image sets to obtain a solid tumor lesion mask erosion training image; performing a pixel-level subtraction operation on the solid tumor lesion mask training image and the solid tumor lesion mask erosion training image to determine the solid tumor lesion edge mask training image corresponding to the solid tumor lesion.

[0102] Specifically, according to the location of the solid tumor lesion in the training image of the solid tumor lesion mask, the second mask value is mapped to the location of the tumor lesion in the training image of the tumor lesion mask, and the first mask value corresponding to the tumor lesion is changed to the second mask value corresponding to the solid tumor lesion within the tumor lesion.

[0103] Specifically, those skilled in the art can configure the first mask value and the second mask value according to actual needs. For example, the first mask value and the second mask value can be configured as 1 and 2, respectively. However, it is worth noting that, unlike the prior art where the preset mask value only represents a region corresponding to a lesion or area to be segmented, the second mask value in the embodiments of this disclosure represents both the first mask value corresponding to the tumor lesion and the second mask value corresponding to the solid tumor lesion within the tumor lesion.

[0104] Specifically, if there is a solid tumor lesion within the tumor lesion, the binary mixed location mask training image includes a first numerical mask corresponding to the tumor lesion and a second numerical mask corresponding to the solid tumor lesion; if there is no solid tumor lesion within the tumor lesion, the binary mixed location mask training image includes a first numerical mask corresponding to the tumor lesion.

[0105] Specifically, the process involves extracting a tumor lesion mask training image with a first mask value corresponding to the tumor lesion in each set of preset location medical training images from the preset location medical training image set, and a solid tumor lesion mask training image with a second mask value corresponding to the solid tumor lesion within the tumor lesion. This includes using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions and solid tumor lesions within the preset location medical training images from a preset number of preset location medical training images, thereby obtaining a tumor lesion mask training image with a first mask value corresponding to the tumor lesion in each set of preset location medical training images, and a solid tumor lesion mask training image with a second mask value corresponding to the solid tumor lesion within the tumor lesion.

[0106] Specifically, the step of training the preset deep learning segmentation network using the binary mixed location mask training image, the solid tumor lesion edge mask training image corresponding to the solid tumor lesion, and the corresponding preset site medical training image to obtain a solid tumor lesion segmentation model includes: obtaining a binary mixed location mask image during the training process of the preset deep learning segmentation network using the binary mixed location mask training image and its corresponding preset site medical training image; using a preset solid tumor lesion edge encoder to encode the solid tumor lesion segmentation mask image corresponding to the solid tumor lesion edge mask image in the binary mixed location mask image and the solid tumor lesion edge mask training image, respectively, to obtain a first edge encoding feature map and a second edge encoding feature map; calculating a first loss of the binary mixed location mask image and its corresponding solid tumor lesion edge mask training image and calculating a second loss of the first edge encoding feature map and the second edge encoding feature map to obtain a solid tumor lesion segmentation model.

[0107] Specifically, those skilled in the art can configure the loss functions corresponding to the first loss and the second loss according to actual needs. For example, the loss functions corresponding to the first loss and the second loss can be configured as one or more of the following loss functions: Cross-Entropy Loss, Dice Loss, Tversky Loss, and Focal Loss.

[0108] Specifically, the binary hybrid location mask image includes: a mask image corresponding to a tumor lesion and a corresponding mask image for a solid tumor lesion within the tumor lesion; wherein the mask values ​​for the tumor lesion and the corresponding mask values ​​for the solid tumor lesion within the tumor lesion are different. Furthermore, the solid tumor lesion segmentation mask image in the binary hybrid location mask image can be determined based on the mask values ​​for the tumor lesion and the corresponding mask values ​​for the solid tumor lesion within the tumor lesion.

[0109] Furthermore, specifically, determining the solid tumor lesion edge mask image includes: performing an erosion operation on the solid tumor lesion segmentation mask image to obtain a solid tumor lesion mask erosion image; performing pixel-level subtraction operations on the solid tumor lesion mask image and the solid tumor lesion mask erosion image to determine the solid tumor lesion edge mask image corresponding to the solid tumor lesion.

[0110] Specifically, the construction of the preset solid tumor lesion edge encoder includes: using the solid tumor lesion edge mask training image corresponding to the solid tumor lesion and its corresponding solid tumor lesion edge training image to construct the preset solid tumor lesion edge encoder.

[0111] Specifically, the step of constructing a preset solid tumor lesion edge encoder using the solid tumor lesion edge mask training image corresponding to the solid tumor lesion and its corresponding solid tumor lesion edge training image includes: extracting the solid tumor lesion edge training image corresponding to each set of preset site medical training images in the preset site medical training image set; and training a preset downsampling convolutional network using the solid tumor lesion edge training image and its corresponding solid tumor lesion edge mask training image to obtain the preset solid tumor lesion edge encoder.

[0112] Specifically, the step of extracting the solid tumor lesion edge training image corresponding to each set of preset site medical training images in the preset site medical training image set includes: performing pixel-level multiplication operations on the solid tumor lesion edge mask training image corresponding to each set of preset site medical training images in the preset site medical training image set and its corresponding preset site medical training image to extract the solid tumor lesion edge training image.

[0113] Specifically, unlike existing technologies, the preset solid tumor lesion edge encoder enables the edges corresponding to solid tumor lesions within the tumor lesion to participate in the training process of the preset deep learning segmentation network using the training image of the two-class hybrid location mask and its corresponding preset site medical training image. This makes the edges corresponding to solid tumor lesions within the tumor lesion controllable and more conducive to the edge segmentation of solid tumor lesions within the tumor lesion.

[0114] Specifically, during the construction of the preset solid tumor lesion edge encoder, a third loss is calculated between the solid tumor lesion edge mask training image corresponding to the solid tumor lesion and its corresponding solid tumor lesion edge training image. Similarly, those skilled in the art can configure the loss function corresponding to the third loss according to actual needs. For example, the loss function corresponding to the third loss can be configured as one or more of the following loss functions: Cross-Entropy Loss, Dice Loss, Tversky Loss, and Focal Loss.

[0115] Specifically, the preset downsampling convolutional network can be configured as one or more of the downsampling convolutional networks corresponding to ResNet, EfficientNet, LeNet, AlexNet, YOLOv10, SRFD / DRFD modules, HWD modules, MobileNet, ShuffleNet, etc.; wherein, those skilled in the art can configure the kernel size, stride of the kernel, pooling and other conventional layers of the above-mentioned preset downsampling convolutional network according to actual needs.

[0116] Specifically, the preset solid tumor lesion edge encoder corresponding to the preset downsampling convolutional network can encode the solid tumor lesion edge mask image corresponding to the solid tumor lesion segmentation mask image in the two-class mixed location mask image and the solid tumor lesion edge mask training image, respectively, thereby achieving one or more objectives such as compressing the spatial size of the solid tumor lesion edge mask image and the solid tumor lesion edge mask training image, expanding the receptive field, and enhancing translation invariance. Specifically, the following features are considered: Compressing spatial dimensions: Reducing the height and width of the feature maps corresponding to the solid tumor lesion edge mask image and the solid tumor lesion edge mask training image reduces subsequent computational load, making the preset solid tumor lesion edge encoder more efficient; Expanding the receptive field: As the size of the feature maps corresponding to the solid tumor lesion edge mask image and the solid tumor lesion edge mask training image decreases, neurons in subsequent layers can cover a wider input region, helping to capture the global features corresponding to the solid tumor lesion edge mask image and the solid tumor lesion edge mask training image; Enhancing translation invariance: The encoder is more robust to small displacements or deformations of the solid tumor lesion edge mask image and the solid tumor lesion edge mask training image, improving the generalization ability of the preset solid tumor lesion edge encoder.

[0117] For example, the pre-defined downsampling convolutional network can be configured as a hierarchical feature extraction structure or a gradually expanding receptive field structure. For the hierarchical feature extraction structure, it typically consists of multiple convolutional layers and pooling layers stacked alternately. With each downsampling operation, the resolution of the feature maps corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask decreases, but the number of channels may increase to retain richer semantic information. For the gradually expanding receptive field structure, with each downsampling operation on the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask, the input region that the deep neurons of the network can "see" becomes larger and larger, which helps to gradually transition from the local features corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask to the global semantic understanding corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask.

[0118] For example, by setting the stride of the convolutional kernel to be greater than 1 (e.g., stride=2), the convolutional kernel slides in jumps across the input feature maps corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask, directly reducing the size of the output feature maps corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask. For example, if the input size is 5×5, the convolutional kernel size is 3×3, and the stride is 2, the output size corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask will become 2×2. Common max pooling and average pooling downsample the feature maps corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask by selecting the maximum or average value of the local region corresponding to the solid tumor lesion edge mask image and the training image of the solid tumor lesion edge mask. For example, 2×2 max pooling will halve the size of the feature map corresponding to the above solid tumor lesion edge mask image and the solid tumor lesion edge mask training image.

[0119] Specifically, before segmenting the medical image of the preset location using a deep learning-based solid tumor lesion segmentation model, the process includes: constructing tumor lesions for each set of medical training images of the preset location in a set of medical training images of the preset location, and binary mixed location mask training images corresponding to solid tumor lesions within the tumor lesions; training the preset deep learning segmentation network using the binary mixed location mask training images, the solid tumor lesion edge mask training images corresponding to the solid tumor lesions, and the corresponding medical training images of the preset location to obtain a solid tumor lesion segmentation model; and segmenting solid tumor lesions in the medical image of the preset location to be segmented (the medical image of the preset location) based on the solid tumor lesion segmentation model.

[0120] Specifically, based on the tumor lesion mask image corresponding to the medical image of the preset location, the tumor lesion image corresponding to the tumor lesion mask image is extracted from the medical image of the preset location; based on the tumor lesion image, one or more of the following are determined: the long diameter of the solid component of the tumor lesion, the ratio of the solid component of the tumor lesion to the tumor lesion in which it is located, and the air bronchogram sign of the tumor lesion.

[0121] Specifically, detecting the spinous process sign of the tumor lesion includes: calculating the ratio of the long axis to the short axis of each tumor lesion in a masked image of the tumor lesion corresponding to a medical image of a preset location; wherein the medical image of the preset location is configured as a two-dimensional medical image; if the ratio of the long axis to the short axis is greater than or equal to a first preset ratio of the long axis to the short axis, then the corresponding tumor lesion is determined to have a spinous process sign. This addresses at least one of the technical problems currently present, such as the excessive subjectivity in manually determining whether a spinous process sign exists at the edge of a tumor lesion, the huge workload for doctors due to the lack of intelligent determination of whether a spinous process sign exists at the edge of a tumor lesion, and the inability to effectively evaluate the subsequent prediction of radiotherapy efficacy.

[0122] This disclosure also proposes a method for detecting a spike-like feature, comprising: calculating multiple ratios of the long and short diameters of each three-dimensional tumor lesion in a mask image of a tumor lesion corresponding to a preset medical image; if any one of the ratios is greater than or equal to a first preset ratio, then the corresponding tumor lesion is determined to have a spike-like feature; wherein the preset medical image is configured as a three-dimensional medical image. This addresses at least one of the technical problems currently present, including the need for manual determination of whether a spike-like feature exists at the edge of a tumor lesion, which is too subjective; the lack of intelligent determination of whether a spike-like feature exists at the edge of a tumor lesion, resulting in a huge workload for doctors; and the inability to effectively evaluate the subsequent prediction of radiotherapy efficacy.

[0123] This disclosure also proposes a method for detecting a spike-like feature, comprising: determining whether a medical image of a preset location is a two-dimensional or three-dimensional medical image; if the medical image of the preset location is a two-dimensional medical image, calculating the ratio of the long axis to the short axis of each tumor lesion in the tumor lesion mask image corresponding to the preset location medical image; if the ratio of the long axis to the short axis is greater than or equal to a first preset ratio of the long axis to the short axis, determining that the corresponding tumor lesion has a spike-like feature; otherwise, determining that the corresponding tumor lesion does not have a spike-like feature; if the medical image of the preset location is a three-dimensional medical image, calculating multiple ratios of the long axis to the short axis of each three-dimensional tumor lesion in the tumor lesion mask image corresponding to the preset location medical image; if any one of the ratios of the long axis to the short axis is greater than or equal to the first preset ratio of the long axis to the short axis, determining that the corresponding tumor lesion has a spike-like feature.

[0124] In this embodiment of the disclosure, determining the major axis corresponding to each three-dimensional tumor lesion includes: performing three-dimensional reconstruction on each tumor lesion mask in the first tumor lesion mask image to obtain the three-dimensional mask tumor lesion corresponding to each tumor lesion mask; performing edge detection on the three-dimensional mask tumor lesion of each tumor lesion mask to obtain the three-dimensional tumor lesion mask edge; selecting any three-dimensional edge pixel point in the three-dimensional tumor lesion mask edge; calculating multiple three-dimensional edge pixel point distances between the arbitrary three-dimensional edge pixel point and other three-dimensional edge pixel points; selecting the maximum three-dimensional edge pixel point distance from the multiple three-dimensional edge pixel point distances; and configuring the maximum three-dimensional edge pixel point distance as the major axis corresponding to each three-dimensional tumor lesion.

[0125] In this embodiment of the disclosure, determining the short axis corresponding to each tumor lesion includes: performing three-dimensional reconstruction on each tumor lesion mask in the first tumor lesion mask image to obtain a three-dimensional mask tumor lesion corresponding to each tumor lesion mask; performing edge detection on the three-dimensional mask tumor lesion of each tumor lesion mask to obtain the three-dimensional tumor lesion mask edge; selecting any three-dimensional edge pixel in the three-dimensional tumor lesion mask edge; calculating multiple three-dimensional edge pixel distances between the arbitrary three-dimensional edge pixel and other three-dimensional edge pixel; selecting the minimum three-dimensional edge pixel distance from the multiple three-dimensional edge pixel distances; configuring the minimum three-dimensional edge pixel distance as the long axis corresponding to each three-dimensional lesion; if the ratio of the long axis to the short axis is less than the first preset ratio of the long axis to the short axis, then it is determined that the corresponding tumor lesion does not have a spike sign.

[0126] In this embodiment, determining the tumor lesion mask image corresponding to the preset location medical image includes: training a first preset deep learning segmentation network using a preset number of preset location medical training images and their corresponding tumor lesion mask training images to obtain a preset tumor lesion segmentation model; segmenting the tumor lesion in the preset location medical image using the preset tumor lesion segmentation model to obtain the corresponding tumor lesion mask image; segmenting the artery using the preset artery segmentation model to obtain the corresponding artery mask image; and / or, configuring the tumor lesion in the preset location medical image as a tumor lesion to be treated with chemotherapy or a tumor lesion to be operated on. Wherein, the mask value corresponding to each tumor lesion region in the tumor lesion mask image and the tumor lesion mask training image is configured as 1, and the mask value corresponding to non-tumor lesion regions is configured as 0.

[0127] Specifically, before training the first preset deep learning segmentation network using a preset number of preset site medical training images and their corresponding tumor lesion mask training images, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions on the preset number of preset site medical training images to obtain corresponding tumor lesion mask training images.

[0128] In this embodiment of the disclosure, if the tumor lesion has a spinous process, then based on the tumor lesion mask image and arterial vessel mask image corresponding to the preset site medical image, it is determined whether the spinous process corresponding to the tumor lesion is adjacent to an arterial vessel; if the spinous process corresponding to the tumor lesion is adjacent to an arterial vessel, then the spinous process is determined to be a dangerous spinous process.

[0129] In this embodiment of the disclosure, if the tumor lesion has a spinous process, then based on the tumor lesion mask and arterial vessel mask image corresponding to the preset site medical image, it is determined whether the spinous process corresponding to the tumor lesion is adjacent to an arterial vessel; if the spinous process corresponding to the tumor lesion is adjacent to an arterial vessel, then the spinous process is determined to be a dangerous spinous process.

[0130] In this embodiment of the disclosure, determining whether the spinous process corresponding to the tumor lesion is adjacent to an artery based on a tumor lesion mask image and an artery mask image corresponding to a preset medical image includes: extracting multiple spinous process position coordinates corresponding to multiple spinous processes in the tumor lesion mask image and artery mask edge position coordinates corresponding to the artery mask edge in the artery mask image; calculating multiple distances between the multiple spinous process position coordinates and the artery mask edge position coordinates; if the minimum distance among the multiple distances is less than a preset distance, then it is determined that the spinous process corresponding to the tumor lesion is adjacent to an artery.

[0131] In this embodiment of the disclosure, determining whether the spinous process corresponding to the tumor lesion is adjacent to an artery based on the tumor lesion mask and artery mask image corresponding to a preset medical image includes: extracting multiple spinous process position coordinates corresponding to multiple spinous processes and artery mask edge position coordinates corresponding to the artery mask edge in the tumor lesion mask and artery mask images; calculating multiple distances between the multiple spinous process position coordinates and the artery mask edge position coordinates respectively; if the minimum distance among the multiple distances is less than a preset distance, then it is determined that the spinous process corresponding to the tumor lesion is adjacent to an artery.

[0132] Specifically, determining the thorn includes: detecting the edges of the tumor lesion mask in the tumor lesion mask image to obtain the tumor lesion mask edge line; calculating the first pixel distance and the second pixel distance between any tumor lesion mask edge pixel on the tumor lesion mask edge line and its adjacent tumor lesion mask edge pixels; if the first pixel distance and the second pixel distance are greater than a set pixel distance, then the arbitrary tumor lesion mask edge pixel is configured as a thorn.

[0133] Specifically, determining whether the location of the tumor lesion to be treated or operated on, based on the tumor lesion mask image and the artery mask image, is adjacent to an artery includes: calculating multiple positional distances between the artery edge mask position coordinates corresponding to the artery mask image and the tumor lesion edge mask position coordinates corresponding to the tumor lesion mask image; if the minimum positional distance among the multiple positional distances is less than or equal to a preset positional distance, then the tumor lesion is determined to be adjacent to an artery; otherwise, the tumor lesion is determined to be far from an artery. The artery edge mask position coordinates and the tumor lesion edge mask position coordinates are respectively configured as the position coordinates of the artery edge in the artery edge mask image corresponding to the artery mask image and the position coordinates of the tumor edge mask in the tumor lesion edge mask image corresponding to the tumor lesion mask image.

[0134] Specifically, determining whether the location of the tumor lesion to be treated or operated on, based on the tumor lesion mask and artery mask images, is adjacent to an artery includes: calculating multiple positional distances between the artery edge mask position coordinates corresponding to the tumor lesion mask and the artery mask position coordinates; if the minimum positional distance among the multiple positional distances is less than or equal to a preset positional distance, then the tumor lesion is determined to be adjacent to an artery; otherwise, the tumor lesion is determined to be far from an artery. The artery edge mask position coordinates and the tumor lesion edge mask position coordinates are respectively configured as the position coordinates of the artery edge in the artery edge mask and the position coordinates of the tumor lesion edge in the tumor lesion mask. Those skilled in the art can configure the preset positional distance according to actual needs. For example, the preset positional distance can be any value between 0 and 10 pixels or other values.

[0135] Specifically, the location of tumor lesions and their proximity to arteries are of significant clinical importance for indicating disease type, aiding in differential diagnosis, and influencing subsequent chemotherapy and surgery. It is necessary to propose a technical solution for determining tumor lesion location and for intelligent processing of medical images to address at least one of the following technical problems: the current reliance on manual determination of tumor lesion location and proximity to arteries leads to excessive subjectivity; the lack of intelligent determination results in a heavy workload for doctors; and it prevents effective evaluation of subsequent radiotherapy efficacy predictions.

[0136] Specifically, the step of determining the tumor lesion mask image and arterial vessel mask image corresponding to the tumor lesion to be treated or the tumor lesion to be operated on using medical images of preset locations includes: training a first preset deep learning segmentation network using a preset number of preset location medical training images and their corresponding tumor lesion mask training images to obtain a preset tumor lesion segmentation model; training a second preset deep learning segmentation network using a preset number of preset location medical training images and their corresponding arterial vessel mask training images to obtain a preset arterial vessel segmentation model; segmenting the tumor lesion to be treated or the tumor lesion to be operated on in the preset location medical images using the preset tumor lesion segmentation model to obtain the corresponding tumor lesion mask image; and segmenting the arterial vessels using the preset arterial vessel segmentation model to obtain the corresponding arterial vessel mask image.

[0137] Specifically, the step of determining the tumor lesion mask and arterial vessel mask images corresponding to the tumor lesions to be treated or the tumor lesions to be operated on using medical images of preset locations includes: training a preset deep learning segmentation network using a preset number of preset location medical training images and their corresponding tumor lesion mask and arterial vessel mask training images to obtain a preset tumor lesion and arterial vessel segmentation model; and segmenting the tumor lesion to be treated or the tumor lesion to be operated on and the arterial vessels using the preset tumor lesion and arterial vessel segmentation model to obtain the corresponding tumor lesion mask and arterial vessel mask images.

[0138] Specifically, the mask value corresponding to each tumor lesion region in the tumor lesion mask image and the tumor lesion mask training image is configured as 1, and the mask value corresponding to the non-tumor lesion region is configured as 0; similarly, the mask value corresponding to the artery vessels in the artery vessel mask image and the artery vessel mask training image is configured as 1, and the mask value corresponding to the non-artery vessel region is configured as 0.

[0139] Specifically, in the tumor lesion mask and artery mask images and the tumor lesion mask and artery mask training images, the mask value corresponding to each tumor lesion region is configured as 1, the mask value corresponding to the artery is configured as 2, and the mask value corresponding to non-tumor lesion regions and non-artery regions is configured as 0. Before training the first preset deep learning segmentation network using a preset number of preset site medical training images and their corresponding tumor lesion mask training images, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions in the preset number of preset site medical training images to obtain corresponding tumor lesion mask training images. Similarly, before training the second preset deep learning segmentation network using a preset number of preset site medical training images and their corresponding artery mask training images, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate arteries in the preset number of preset site medical training images to obtain corresponding artery mask training images.

[0140] Specifically, before training the preset deep learning segmentation network using a preset number of preset site medical training images and their corresponding tumor lesion masks and artery mask training images, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions and arteries on the preset number of preset site medical training images to obtain corresponding tumor lesion mask and artery mask training images.

[0141] Specifically, the preset deep learning segmentation network, the first preset deep learning segmentation network, and the second preset deep learning segmentation network are configured as one or more deep learning segmentation networks such as Fully Convolutional Network (FCN), U-Net, DeepLab, and Pyramid Spatial Pooling Network (PSPNet).

[0142] Specifically, if the edge of the tumor lesion has a spiky sign, then determining whether the tumor lesion in the preset location medical image is a tumor lesion to be treated with chemotherapy or a tumor lesion to be operated on includes: detecting whether the boundary of the tumor lesion in the tumor lesion mask image corresponding to the preset location medical image is blurred; if the boundary of the tumor lesion in the tumor lesion mask image is blurred, then the tumor lesion is configured as a tumor lesion to be treated with chemotherapy or a tumor lesion to be operated on.

[0143] Specifically, if the tumor lesion has a spiky edge, then determining whether the tumor lesion in the preset location medical image is a tumor lesion to be treated with chemotherapy or a tumor lesion to be operated on includes: detecting whether the boundary of the tumor lesion in the preset location medical image and the arterial mask image is blurred and whether there is a spiky edge on the tumor lesion; if the boundary of the tumor lesion in the tumor lesion and the arterial mask image is blurred and there is a spiky edge on the tumor lesion, then the tumor lesion is configured as a tumor lesion to be treated with chemotherapy or a tumor lesion to be operated on.

[0144] Specifically, determining whether the boundary of the tumor lesion is blurred includes: mapping the location coordinates of the tumor lesion in the tumor lesion masking image to a medical image of a preset location; calculating multiple gradients corresponding to the location coordinates of the tumor lesion in the medical image of the preset location; if the gradient magnitude value corresponding to the multiple gradients is less than a preset gradient magnitude value, then the boundary of the tumor lesion where the tumor lesion location is less than the preset gradient magnitude value is determined to be blurred.

[0145] Specifically, if the gradient magnitude value corresponding to the plurality of gradients is greater than or equal to a preset gradient magnitude value, then the boundary of the tumor lesion corresponding to the location of the tumor lesion that is greater than or equal to the preset gradient magnitude value is clearly defined.

[0146] Specifically, before mapping the tumor lesion location coordinates in the tumor lesion masking image to a preset site medical image, the process includes: performing an erosion operation on the tumor lesion mask in the tumor lesion masking image corresponding to the preset site medical image to obtain a tumor lesion mask erosion image; and subtracting the tumor lesion mask erosion image from the tumor lesion masking image to obtain a lesion masking image.

[0147] Specifically, determining the gradient magnitude corresponding to the plurality of gradients includes: squaring the plurality of x-direction gradients corresponding to the coordinates of the tumor lesion location to obtain a plurality of x-direction squared gradients; squaring the plurality of y-direction gradients corresponding to the coordinates of the tumor lesion location to obtain a plurality of y-direction squared gradients; summing the plurality of x-direction squared gradients with their corresponding y-direction squared gradients to obtain a gradient sum corresponding to the coordinates of the plurality of tumor lesions; and taking the square root of the gradient sum corresponding to the coordinates of the plurality of tumor lesions to obtain the gradient magnitude corresponding to the plurality of gradients.

[0148] Specifically, the step of calculating multiple gradients corresponding to the location coordinates of the tumor lesion in the medical image of the preset location includes: determining multiple first position coordinates and multiple second position coordinates on both sides of the location coordinates of the tumor lesion in the medical image of the preset location, corresponding to the x-direction coordinates and the y-direction coordinates respectively; calculating the difference between the first pixel value corresponding to the multiple first position coordinates and the second pixel value corresponding to the multiple second position coordinates respectively to obtain multiple x-direction gradients corresponding to the location coordinates of the tumor lesion; calculating the difference between the third pixel value corresponding to the multiple third position coordinates and the fourth pixel value corresponding to the multiple fourth position coordinates respectively to obtain multiple y-direction gradients corresponding to the location coordinates of the tumor lesion; and configuring the multiple x-direction gradients and their corresponding multiple y-direction gradients as multiple gradients corresponding to the location coordinates of the tumor lesion.

[0149] Specifically, based on the tumor lesion location coordinates, determining multiple first position coordinates and multiple second position coordinates corresponding to both sides of the tumor lesion location coordinates in the preset site medical image in the x-direction includes: using the tumor lesion location coordinates, determining multiple first position coordinates and multiple second position coordinates corresponding to both sides in the x-direction according to preset distances; and / or, based on the tumor lesion location coordinates, determining multiple third position coordinates and multiple fourth position coordinates corresponding to both sides of the tumor lesion location coordinates in the y-direction in the preset site medical image includes: using the tumor lesion location coordinates, determining multiple third position coordinates and multiple fourth position coordinates corresponding to both sides in the x-direction according to preset distances. Those skilled in the art can configure the preset gradient magnitude value according to actual needs. For example, the preset gradient magnitude value can be configured as any value between 2 and 80 or other values.

[0150] Specifically, locating the tumor lesion includes: using a dynamic lung field region division method to divide the medical image of the preset location into lung field regions to obtain a lung field region division network corresponding to each tumor lesion; and locating the lung field anatomical region corresponding to each tumor lesion based on the location of each tumor lesion and its corresponding lung field region division network.

[0151] Specifically, the method of using dynamic lung field region segmentation to segment the lung field region of the medical image of the preset location includes: Step S101: Constructing a set of tumor lesion locations corresponding to all tumor lesion regions using the tumor lesion mask image corresponding to the tumor lesion in the lung field of the medical image of the preset location; Step S102: Traversing the area or volume corresponding to each tumor lesion location in the set of tumor lesion locations to determine the minimum tumor lesion area or minimum tumor lesion volume of all tumor lesion locations; Step S103: Using a preset grid corresponding to the minimum grid area or minimum grid volume of all tumor lesion locations, segmenting the lung field region in the lung field mask image; Step S104: Deleting the tumor lesion location corresponding to the minimum grid area or minimum grid volume in the set of tumor lesion locations to update the set of tumor lesion locations; Step S105: Repeating steps S102-S104 until the lung field region segmentation is completed using all tumor lesion locations in the set of tumor lesion locations. To address at least one of the following technical problems: the excessive subjectivity in tumor lesion location, the lack of intelligent tumor lesion location leading to a huge workload for doctors, and the inability to further predict tumor-induced IgA nephropathy.

[0152] Specifically, each grid in the preset grid predefines a lung field anatomical region; the location of each tumor lesion is configured as the coordinate point corresponding to the medical image of the preset location for each tumor lesion.

[0153] For example, specifically, if the preset site medical image is configured as a two-dimensional preset site medical image, then the coordinate point of each tumor lesion in the preset site medical image is a two-dimensional coordinate (x, y); if the preset site medical image is configured as a three-dimensional preset site medical image, then the coordinate point of each tumor lesion in the preset site medical image is a two-dimensional coordinate (x, y, z).

[0154] Specifically, if the preset site medical image is configured as a two-dimensional preset site medical image, the area corresponding to each tumor lesion location in the tumor lesion location set is traversed to determine the minimum tumor lesion area for all tumor lesion locations; using a preset grid corresponding to the minimum grid area for all tumor lesion locations, the lung field region in the lung field mask image is divided into regions; the tumor lesion location corresponding to the minimum grid area in the tumor lesion location set is deleted to update the tumor lesion location set. If the preset site medical image is configured as a three-dimensional preset site medical image, the area or volume corresponding to each tumor lesion location in the tumor lesion location set is traversed to determine the minimum tumor lesion volume for all tumor lesion locations; using a preset grid corresponding to the minimum grid volume for all tumor lesion locations, the lung field region in the lung field mask image is divided into regions; the tumor lesion location corresponding to the minimum grid volume in the tumor lesion location set is deleted to update the tumor lesion location set.

[0155] For example, specifically, the predefined lung field anatomical regions can be configured as follows: longitudinally, the lungs are divided into upper, middle, and lower lung fields, with the lower borders of the 2nd and 4th anterior ribs as boundaries; and laterally, each lung field is vertically divided into an inner zone (near the mediastinum), a middle zone, and an outer zone (near the chest wall). The lung field anatomical regions can also be configured as the apical region corresponding to the area surrounded by the outer border of the 1st rib, and the subclavian region corresponding to the area between the lower border of the clavicle and the lower border of the 2nd anterior rib. The predefined lung field anatomical regions can also be configured as the upper, middle, and lower lobes of the right lung and the upper and lower lobes of the left lung; wherein the upper lobe of the left lung may include the lingular segment. The predefined lung field anatomical regions can be configured as lung segments corresponding to independent distribution units of bronchi and blood vessels, each lung segment having independent bronchial and vascular bundles; the upper lobe corresponding to the right lung may include the apical segment, posterior segment, and anterior segment; the middle lobe corresponding to the right lung may include the lateral segment and medial segment; the lower lobe corresponding to the right lung may include the dorsal segment, medial basal segment, anterior basal segment, lateral basal segment, and posterior basal segment; the upper lobe corresponding to the left lung may include the apical-posterior segment, anterior segment, superior lingular segment, and inferior lingular segment; the lower lobe corresponding to the left lung may include the dorsal segment, medial anterior basal segment, lateral basal segment, and posterior basal segment.

[0156] Specifically, the predefined lung field anatomical regions can be arbitrarily combined without violating the anatomical structure of the lung field.

[0157] Specifically, determining the tumor lesion mask image corresponding to the tumor lesion in the lung field of the medical image at the preset location includes: segmenting the tumor lesion in the medical image at the preset location using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network, and determining the tumor lesion mask image of the medical image at the preset location.

[0158] Specifically, before segmenting the medical images of the preset locations using a preset segmentation model constructed based on a deep learning segmentation network, the process includes: training the preset deep learning segmentation network using a preset number of medical images of the preset locations and their corresponding tumor lesion mask training images to obtain a preset tumor lesion segmentation model constructed based on the deep learning segmentation network. Specifically, before training the preset deep learning segmentation network, the process includes: using segmentation annotation software to delineate tumor lesions on the preset number of medical images of the preset locations to obtain corresponding tumor lesion mask training images.

[0159] Specifically, before training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain a preset segmentation model constructed based on the deep learning segmentation network, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate tumor lesions on the preset number of medical images of preset locations, obtaining corresponding tumor lesion mask training images. Those skilled in the art can configure the preset number according to actual needs. For example, the preset number can be configured as 300-500.

[0160] Specifically, it also includes: determining whether each tumor lesion is adjacent to a pulmonary artery.

[0161] Specifically, determining whether each tumor lesion is adjacent to a pulmonary artery includes: calculating multiple positional distances between the pulmonary artery edge mask position coordinates corresponding to the pulmonary artery mask image of the preset site medical image and the tumor lesion edge mask position coordinates corresponding to the tumor lesion mask image; if the minimum positional distance among the multiple positional distances is less than or equal to the preset positional distance, then the tumor lesion is determined to be adjacent to a pulmonary artery; otherwise, the tumor lesion is determined to be far from a pulmonary artery.

[0162] Specifically, determining the pulmonary artery mask image of the medical image of the preset location includes: using a preset pulmonary artery segmentation model constructed based on a deep learning segmentation network to segment the pulmonary artery of the medical image of the preset location, and determining the pulmonary artery mask image of the medical image of the preset location.

[0163] Specifically, before performing pulmonary artery segmentation on the medical images of the preset locations using a preset segmentation model constructed based on a deep learning segmentation network, the process includes: training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding pulmonary artery mask training images to obtain a preset pulmonary artery segmentation model constructed based on the deep learning segmentation network. Specifically, before training the preset deep learning segmentation network, the process includes: using segmentation annotation software to delineate the pulmonary artery vessels in a preset number of preset medical images to obtain corresponding pulmonary artery vessel mask training images.

[0164] Specifically, the position coordinates of the pulmonary artery vessel edge mask and the position coordinates of the tumor lesion edge mask are respectively configured as the position coordinates of the pulmonary artery vessel edge in the pulmonary artery vessel edge mask image corresponding to the pulmonary artery vessel mask image and the position coordinates of the tumor lesion edge mask in the tumor lesion edge mask image corresponding to the tumor lesion mask image.

[0165] Specifically, before training a preset deep learning segmentation network using a preset number of preset site medical images and their corresponding pulmonary artery vascular mask training images to obtain a preset segmentation model constructed based on the deep learning segmentation network, the process includes: using segmentation annotation software CasiaLabeler or Labelme to delineate the pulmonary arteries in the preset number of preset site medical images, obtaining corresponding pulmonary artery vascular mask training images. Those skilled in the art can configure the preset number according to actual needs. For example, the preset number can be configured as 300-500.

[0166] Similarly, specifically, the preset deep learning segmentation network is configured as one or more deep learning segmentation networks such as Fully Convolutional Network (FCN), U-Net, DeepLab, and Pyramid Spatial Pooling Network (PSPNet).

[0167] Specifically, those skilled in the art can configure the preset position distance according to actual needs. For example, the preset position distance can be any value between 0 and 10 pixels or other values.

[0168] Specifically, the location of tumor lesions and their proximity to pulmonary arteries are of significant clinical importance for indicating disease type, assisting in differential diagnosis, and subsequent chemotherapy and surgery. This aims to address at least one of the following technical problems: the current reliance on manual determination of tumor lesion location and proximity to pulmonary arteries leads to excessive subjectivity, a lack of intelligent determination of tumor lesion location and proximity to pulmonary arteries, resulting in a huge workload for doctors and hindering further surgical procedures and prediction of tumor-induced IgA nephropathy.

[0169] Specifically, determining the lobulation sign of the tumor lesion includes: performing an erosion operation on the tumor lesion mask image corresponding to each tumor lesion in the medical image of the preset location to obtain a eroded tumor lesion mask image; subtracting the corresponding eroded tumor lesion mask image from the tumor lesion mask image of each tumor lesion to obtain a processed tumor lesion mask image corresponding to each tumor lesion; judging the morphology of the tumor lesion in the processed tumor lesion mask image corresponding to each tumor lesion; if the morphology of the tumor lesion is irregular, determining whether the lobulation sign of the tumor lesion exists in the processed tumor lesion mask image corresponding to each tumor lesion.

[0170] Specifically, determining the lobulation sign of tumor lesions includes: determining the tumor lesion mask processing image corresponding to each tumor lesion in the tumor lesion mask image of the medical image of the preset location; judging the morphology of the tumor lesion in the tumor lesion mask processing image corresponding to each tumor lesion; if the morphology of the tumor lesion is irregular, determining whether the tumor lesion lobulation sign exists in the tumor lesion mask processing image corresponding to each tumor lesion.

[0171] Specifically, the step of judging the morphology of the tumor lesion in the tumor lesion mask processing image corresponding to each tumor lesion includes: extracting multiple tumor lesion mask position coordinates corresponding to the tumor lesion mask line in the tumor lesion mask processing image corresponding to each tumor lesion; fitting or reconstructing the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion corresponding to each tumor lesion; and determining whether the morphology of the tumor lesion is regular based on the morphology of the tumor lesion corresponding to each tumor lesion.

[0172] Specifically, the step of processing the image based on the tumor lesion mask corresponding to each tumor lesion to determine whether there is a lobulation sign of the tumor lesion includes: extracting the tumor lesion mask line corresponding to each tumor lesion; if the shape of the tumor lesion mask line is an arc-shaped convex connection, then the corresponding tumor lesion has a lobulation sign; otherwise, there is no lobulation sign of the tumor lesion.

[0173] Specifically, the step of processing the image based on the tumor lesion mask corresponding to each tumor lesion to determine whether there is a lobulation sign of the tumor lesion further includes: segmenting the tumor lesion mask line according to its concave part to obtain multiple arc-shaped convex line segments; determining whether the arc-shaped convex is a tip based on the slope corresponding to each pixel in the multiple arc-shaped convex line segments; if it is a tip, then the corresponding tumor lesion does not have a lobulation sign of the tumor lesion and the tumor lesion has a spiky sign; if there is no tip, then the corresponding tumor lesion has a lobulation sign of the tumor lesion.

[0174] Specifically, determining whether an arc-shaped protrusion is a tip based on the slope corresponding to each pixel in the multiple arc-shaped protrusion segments includes: mapping each of the multiple arc-shaped protrusion segments to a Cartesian coordinate system, with the coordinate point corresponding to the arc-shaped protrusion of each segment being the highest pixel in the Cartesian coordinate system; calculating the left and right derivatives corresponding to the highest pixel; if the difference between the left and right derivatives is less than or equal to a set difference, then the arc-shaped protrusion is determined to be a tip; otherwise, the arc-shaped protrusion is determined to be non-tip. Those skilled in the art can configure the set difference as needed. For example, the set difference can be configured as 1, 2, or other possible values.

[0175] Specifically, the step of fitting or reconstructing the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion includes: under the optimal criterion, approximating multiple preset closed curve equations or multiple preset closed surface equations using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion corresponding to each tumor lesion.

[0176] Specifically, under the optimal criterion, the step of approximating a preset closed curve equation or a preset closed surface equation using the multiple tumor lesion mask position coordinates corresponding to each tumor lesion to determine the morphology of the tumor lesion corresponding to each tumor lesion includes: under the optimal criterion, minimizing the error or absolute residual between the approximation equation and the multiple preset closed curve equation or multiple preset closed surface equation when the multiple tumor lesion mask position coordinates corresponding to each tumor lesion approximate the multiple preset closed curve equation or multiple preset closed surface equation, thereby determining the morphology of the tumor lesion corresponding to each tumor lesion.

[0177] Specifically, before processing the tumor lesion mask image corresponding to the tumor lesion in the tumor lesion mask image of the medical image of the preset location, the process includes: using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network to segment the tumor lesion in the medical image of the preset location, and determining the tumor lesion mask image of the medical image of the preset location.

[0178] Specifically, before segmenting tumor lesions in medical images of preset locations using a preset segmentation model constructed based on a deep learning segmentation network, the process includes: training a preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain a preset tumor lesion segmentation model constructed based on the deep learning segmentation network.

[0179] Specifically, determining the vacuolar sign of the tumor lesion includes: performing target detection of the vacuolar sign for each tumor lesion in a medical image of a preset location; configuring tumor lesions corresponding to detected vacuolar signs as having the vacuolar sign; extracting the average pixel value of the tumor lesion edge corresponding to tumor lesions for which no vacuolar sign was detected; if the average pixel value corresponding to the tumor lesion for which no vacuolar sign was detected is less than the average pixel value of the corresponding tumor lesion edge, then it is determined that the tumor lesion for which no vacuolar sign was detected has the vacuolar sign; otherwise, it is determined that the tumor lesion for which no vacuolar sign was detected does not have the vacuolar sign. This addresses at least one of the technical problems of existing methods that require manual determination of the presence of vacuolar signs in tumor lesions, which is too subjective and lacks intelligent determination, resulting in a huge workload for doctors and preventing further technical adjustments such as surgery and prediction of IgA nephropathy induced by tumor lesions.

[0180] Specifically, unlike existing technologies, the technical solution proposed in this disclosure detects vacuolar signs in each tumor lesion in a medical image of a preset location. It then uses the average pixel value of the tumor lesion edge corresponding to the lesion where no vacuolar signs were detected to further process the lesions. This effectively solves the technical problem of missed detection of vacuolar signs. Utilizing the average pixel value of the tumor lesion edge corresponding to the lesion where no vacuolar signs were detected improves the robustness of the method for determining whether vacuolar signs exist in tumor lesions where no vacuolar signs were detected. Simultaneously, it avoids at least one of the technical problems associated with existing methods, such as the excessive subjectivity and lack of intelligent determination of vacuolar signs, which leads to a heavy workload for doctors and prevents further surgical procedures or prediction of IgA nephropathy induced by tumor lesions.

[0181] Specifically, the step of performing tumor vacuolation target detection on each tumor lesion in the medical image of the preset location includes: using a deep learning-based tumor vacuolation target detection model to perform tumor vacuolation target detection on each tumor lesion in the medical image of the preset location.

[0182] Specifically, before performing tumor vacuolation target detection on each tumor lesion in the preset site medical image using a deep learning-based tumor lesion vacuolation target detection model, the method includes: training a preset deep learning target detection network to obtain a deep learning-based tumor lesion vacuolation target detection model.

[0183] Specifically, training a preset deep learning target detection network to obtain a deep learning-based tumor lesion vacuolation target detection model includes: training the preset deep learning target detection network using a preset number of preset medical training images of preset locations and their corresponding segmental vacuolation target training images to obtain a deep learning-based tumor lesion vacuolation target detection model.

[0184] Specifically, before training the preset deep learning target detection network, the process includes: using segmentation annotation software to delineate the vacuolar signs of tumor lesions in a preset number of medical images of preset locations, obtaining corresponding segmental vacuolar sign target training images. For example, annotation software such as CasiaLabeler or Labelme can be used to delineate the vacuolar signs of tumor lesions in a preset number of medical images of preset locations, obtaining corresponding segmental vacuolar sign target training images. Those skilled in the art can configure the preset number according to actual needs. For example, the preset number can be configured as 300-500 or other possible values. The preset deep learning target detection network can be configured as a YOLO series or a target detection network based on YOLO.

[0185] Specifically, determining the total number of tumor lesions corresponding to the medical image of the preset location includes: performing tumor lesion segmentation or tumor lesion target detection on the medical image of the preset location to determine the total number of tumor lesions corresponding to the medical image of the preset location; and / or, performing tumor lesion segmentation on the medical image of the preset location includes: performing tumor lesion segmentation on the medical image of the preset location using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network; and / or, performing tumor lesion target detection on the medical image of the preset location includes: performing tumor lesion target detection on the medical image of the preset location using a tumor lesion target detection model based on deep learning.

[0186] Specifically, a preset tumor lesion segmentation model based on a deep learning segmentation network is used to segment the tumor lesions in the medical image of the preset location to obtain the corresponding tumor lesion mask image; the number of tumor lesion masks corresponding to the tumor lesion mask image is counted to determine the total number of tumor lesions corresponding to the medical image of the preset location.

[0187] Specifically, a deep learning-based tumor lesion target detection model is used to detect tumor lesions in the medical images of the preset locations to obtain corresponding tumor lesion detection images; the number of corresponding tumor lesion detection boxes on the tumor lesion detection images is counted to determine the total number of tumor lesions corresponding to the medical images of the preset locations.

[0188] Specifically, before segmenting the medical images of the preset locations using a preset tumor lesion segmentation model constructed based on a deep learning segmentation network, the process includes: training the preset deep learning segmentation network using a preset number of medical images of preset locations and their corresponding tumor lesion mask training images to obtain the preset tumor lesion segmentation model constructed based on the deep learning segmentation network.

[0189] Specifically, before training the preset deep learning segmentation network, the process includes: using segmentation annotation software to delineate tumor lesions on a preset number of preset medical images to obtain corresponding tumor lesion mask training images.

[0190] Specifically, the step of extracting the edge pixel values ​​of tumor lesions for which no vacuolar signs were detected includes: performing edge detection on the mask image of the tumor lesion for which no vacuolar signs were detected to obtain a tumor lesion edge image; performing pixel-level multiplication on the tumor lesion edge image and its corresponding preset site medical image to obtain a tumor lesion edge image; and extracting the pixel values ​​corresponding to the tumor lesion edge in the tumor lesion edge image to obtain the corresponding tumor lesion edge pixel values.

[0191] Specifically, the step of performing edge detection on the tumor lesion mask image corresponding to the tumor lesion for which no vacuolar sign was detected, to obtain a tumor lesion edge image, includes: performing an erosion operation on the tumor lesion mask image corresponding to the tumor lesion for which no vacuolar sign was detected, to obtain a tumor lesion mask erosion image; and subtracting the corresponding tumor lesion mask erosion image from the tumor lesion mask image to obtain a segment edge image.

[0192] Specifically, the step of extracting the average pixel value of the tumor lesion edge corresponding to the tumor lesion for which no tumor vacuolation sign was detected includes: performing edge detection on the tumor lesion mask image corresponding to the tumor lesion for which no tumor vacuolation sign was detected to obtain the tumor lesion edge image; performing pixel-level multiplication on the tumor lesion edge image and its corresponding preset site medical image to obtain the tumor lesion edge image; extracting the pixel value corresponding to the tumor lesion edge in the tumor lesion edge image to obtain the corresponding tumor lesion edge pixel value; and calculating the average value of the tumor lesion edge pixel value to obtain the average pixel value of the tumor lesion edge.

[0193] Specifically, the step of performing edge detection on the tumor lesion mask image corresponding to the tumor lesion for which no vacuolar sign was detected, to obtain a tumor lesion edge image, includes: performing an erosion operation on the tumor lesion mask image corresponding to the tumor lesion for which no vacuolar sign was detected, to obtain a tumor lesion mask erosion image; and subtracting the corresponding tumor lesion mask erosion image from the tumor lesion mask image to obtain a segment edge image.

[0194] Specifically, determining whether a tumor lesion with no detected vacuolation sign exists based on the average pixel value corresponding to the tumor lesion with no detected vacuolation sign and the average pixel value of the corresponding tumor lesion edge includes: if the average pixel value corresponding to the tumor lesion with no detected vacuolation sign is less than the average pixel value of the corresponding tumor lesion edge, then it is determined that the tumor lesion with no detected vacuolation sign exists; otherwise, it is determined that the tumor lesion with no detected vacuolation sign does not exist.

[0195] Specifically, the method for predicting IgA nephropathy induced by tumor lesions further includes: acquiring preset clinical information corresponding to a preset medical image of a preset location at the first time point; training a preset predictor to construct a prediction model for IgA nephropathy induced by tumor lesions using preset quantitative imaging features of the tumor lesion corresponding to the tumor lesion in the medical image of the preset location at the first time point, the preset clinical information, and pathological results corresponding to induced or non-induced IgA nephropathy at a second time point after the first time point; and using the prediction model for IgA nephropathy induced by tumor lesions, the preset quantitative imaging features of the tumor lesion corresponding to the tumor lesion in the medical image of the preset location to be predicted, and the preset clinical information to predict IgA nephropathy induced by tumor lesions.

[0196] Specifically, determining the preset clinical information corresponding to the preset site medical image at the first time point includes: acquiring candidate clinical information corresponding to the preset site medical image at the first time point; and using a univariate logistic regression analysis algorithm, the candidate clinical information, and the corresponding pathological results, performing univariate logistic regression on each candidate clinical information feature to determine the preset clinical information related to the pathological results. Specifically, the clinical information to be selected is configured as one or more of the following: age, gender, smoking history, and history of lung cancer. Prior to training the preset predictor, the process includes: standardizing or normalizing the numerical clinical information to be selected, and encoding the categorized clinical information to be selected.

[0197] Specifically, for example, before training the preset predictor, the age corresponding to the clinical information to be selected is standardized or normalized, and one or more of the following are encoded: gender, smoking history, and whether or not there is a history of lung cancer.

[0198] Specifically, for example, encoding processing is performed on one or more of the following clinical information corresponding to the categorized selection: gender, smoking history, and presence or absence of a history of lung cancer. This includes: configuring a first value for males; configuring a second value for females; configuring a first value for those with a smoking history; configuring a second value for those without a smoking history; configuring a first value for those with a history of lung cancer; and configuring a second value for those with a history of lung cancer. Those skilled in the art can configure the first and second values ​​according to actual needs. For example, the first and second values ​​can be configured as 1 and 0, respectively.

[0199] Current research has confirmed that periodontitis plays a synergistic and promoting role in the occurrence and development of many systemic diseases. Studies have shown that periodontitis and diabetes mutually promote each other, and that blood sugar control requires the simultaneous management of periodontitis. Periodontitis is closely related to cardiovascular and cerebrovascular diseases, participates in lipid metabolism processes, and is closely associated with rheumatoid arthritis, premature birth in pregnant women, blood diseases, and gastrointestinal tumors. Chronic periodontitis is also closely related to low albumin levels in maintenance hemodialysis patients; therefore, periodontitis can be considered a potential indicator of chronic diseases. Thus, periodontitis participates in the occurrence and development of many diseases, and studying the relationship between periodontitis and IgA nephropathy is crucial for early warning of kidney disease based on the association between IgA and periodontal plaque.

[0200] Dental plaque is mainly distributed on the tooth surface, especially in hard-to-clean areas such as the gum line, between teeth, pits and fissures (the depressions on the occlusal surface of teeth), and the complex grooves of posterior teeth. Given the close relationship between IgA nephropathy and mucosal immunity, the health of the periodontal mucosa is closely related to the periodontal microbiota. However, when collecting dental plaque from patients with IgA nephropathy and chronic severe periodontitis to extract plaque DNA, there is currently a lack of effective quantitative analysis methods using oral X-ray images to determine the probe insertion depth for plaque collection. If the insertion depth between the tooth and adjacent periodontal tissues on one side does not reach the preset probe depth, forcibly inserting the probe into this space can damage the periodontal tissues and hinder plaque collection, thus affecting kidney disease early warning decisions related to periodontal plaque.

[0201] Based on the above, the method further includes: extracting the tooth boundary line within the periodontal tissue corresponding to the preset side of the tooth where the tooth probe is inserted from the oral X-ray image to be analyzed; and estimating the probe insertion depth to determine the gap between the preset side of the tooth and its adjacent periodontal tissue based on the tooth boundary line within the periodontal tissue. This aims to solve at least one of the technical problems such as causing damage to periodontal tissue, hindering plaque collection, and affecting kidney disease early warning decisions associated with periodontal plaque.

[0202] Specifically, the tooth boundary line within the periodontal tissue on a predetermined side of the tooth is extracted from the oral X-ray image to be analyzed from a predetermined orientation. The predetermined orientation is configured as a lateral cranial view or a frontal cranial view. The periodontal tissue may include: the gingiva, periodontal ligament, cementum, and alveolar bone. The tooth is enclosed within the periodontal tissue, and the tooth boundary can form a tooth boundary line within the periodontal tissue that includes the tooth root (generally the tooth root boundary line) or a tooth boundary line that includes the tooth root and the portion of the crown connected to the tooth root.

[0203] Specifically, unlike computed tomography (CT) technology, due to the principle of X-ray radiation technology, when using an X-ray camera to photograph the oral cavity, the oral anatomy characteristic of the periodontal tissue surrounding the tooth root will create an overlapping area of ​​periodontal tissue and tooth root on the oral X-ray image. In other words, on the oral X-ray image, tooth boundary lines will be formed on both sides of the tooth within the periodontal tissue.

[0204] Specifically, the preset side of the tooth can be configured as one or more of the left or right side between adjacent teeth, the front side of the tooth, and the back side of the tooth. When the preset side of the tooth can be configured as the left or right side between adjacent teeth, it is necessary to determine multiple gaps between multiple teeth based on the oral X-ray image to be analyzed, and then determine the preset side of the tooth based on the multiple gaps between the multiple teeth.

[0205] Specifically, determining the preset side of the tooth includes: determining multiple gaps between multiple teeth based on the oral X-ray image to be analyzed; if the multiple gaps are larger than a preset gap, then the side of the tooth corresponding to the larger preset gap is determined as the preset side of the tooth. Specifically, when there is no overlapping area between the tooth mask corresponding to each tooth and the tooth mask corresponding to its adjacent teeth, and there is no overlapping area corresponding to the first position tooth and the second position tooth adjacent to the first position tooth, the maximum distance between the edge line of the first tooth mask corresponding to the first position tooth and the edge line of the second tooth mask corresponding to the second position tooth is calculated. Specifically, determining the multiple gaps between multiple teeth based on the oral X-ray image to be analyzed includes: performing tooth segmentation on the oral X-ray image to be analyzed to obtain a tooth mask image; determining multiple gap values ​​between multiple teeth based on the edge of the tooth mask corresponding to each tooth in the tooth mask image.

[0206] Specifically, determining multiple gap values ​​between multiple teeth based on the tooth mask corresponding to each tooth in the tooth mask image includes: determining whether there is an overlapping region between the tooth mask corresponding to each tooth and the tooth mask corresponding to its adjacent teeth; if the overlapping region exists, then there is no gap between the teeth corresponding to the overlapping region; if the overlapping region does not exist, then the maximum distance between the edge lines of the tooth mask corresponding to the teeth that do not have the overlapping region is calculated; and the maximum distance is determined as the multiple gap values ​​between multiple teeth.

[0207] Specifically, before determining the tooth mask corresponding to each tooth in the oral X-ray image to be analyzed, the process includes: segmenting the teeth in the oral X-ray image to be analyzed using a pre-defined tooth segmentation model based on deep learning to obtain a tooth mask image. Specifically, constructing the pre-defined tooth segmentation model includes: training the pre-defined deep learning segmentation model using a pre-defined number of oral X-ray training images and their corresponding tooth mask training images to construct the pre-defined segmentation model. Specifically, determining the tooth mask training images includes: using segmentation annotation software to delineate the teeth in each of the pre-defined number of oral X-ray training images to obtain the corresponding tooth mask training image. The segmentation annotation software can be configured as CasiaLabeler, Labelme, or other possible annotation software. The pre-defined deep learning segmentation network can be configured as one or more deep learning segmentation networks such as Fully Convolutional Network (FCN), U-Net, DeepLab, and Pyramid Spatial Pooling Network (PSPNet).

[0208] Specifically, determining the preset side of the tooth further includes: if the first gap on both sides of the tooth is larger than the preset gap and the second gap is larger than the preset gap, then extract the first overlapping area and the second overlapping area formed by the adjacent periodontal tissues on both sides of the tooth in the oral X-ray image to be analyzed; and determine the preset side of the tooth based on the first overlapping area and the second overlapping area.

[0209] Specifically, determining the preset side of the tooth based on the first overlapping area and the second overlapping area includes: if the first overlapping area is smaller than the second overlapping area, then the side between the tooth and periodontal tissue corresponding to the first overlapping area is determined as the preset side of the tooth; otherwise, the other side between the tooth and periodontal tissue corresponding to the second overlapping area is determined as the preset side of the tooth.

[0210] Specifically, extracting the first and second overlapping areas formed by adjacent periodontal tissues on both sides of the tooth in the oral X-ray image to be analyzed includes: determining a tooth area average line based on a tooth mask image corresponding to the tooth; determining the area formed by the tooth boundary line within the periodontal tissue corresponding to one side of the tooth area average line and the area formed by the periodontal tissue as the first overlapping area; and determining the area formed by the tooth boundary line within the periodontal tissue corresponding to the other side of the tooth area average line and the area formed by the periodontal tissue as the second overlapping area. The direction of the tooth area average line is either from the root to the crown or from the crown to the root.

[0211] Specifically, determining the preset side of the tooth further includes: if the first gap on both sides of the tooth is larger than the preset gap and the second gap is larger than the preset gap, then based on the preset tooth probe insertion side discrimination model, the preset side of the tooth is determined using the first overlap area and the second overlap area formed by the adjacent periodontal tissues on both sides of the tooth in the oral X-ray image to be analyzed. Constructing the preset tooth probe insertion side discrimination model includes: extracting the first overlap area and the second overlap area for training from a preset number of oral X-ray training images; training a preset classifier using the first overlap area, the second overlap area for training, and their corresponding known probe insertion positions to construct the preset probe position discrimination model.

[0212] Specifically, the step of training a preset classifier using the first overlap area, the second overlap area, and their corresponding known probe insertion positions to construct the preset probe position discrimination model includes: determining the preset probe position discrimination model corresponding to each preset classifier in the preset classifier pool corresponding to the first overlap area and the second overlap area used for training, and the corresponding sub-preset probe position discrimination model corresponding to each preset classifier in the sub-training set of the first overlap area and the second overlap area used for training; determining a sub-level probability value set on the sub-validation set using the sub-preset probe position discrimination model and the sub-validation set of the first overlap area and the second overlap area used for training; determining a level probability value set on the validation set using the preset probe position discrimination model and the validation set corresponding to the first overlap area and the second overlap area used for training; and optimizing the level probability value set using the fusion weights corresponding to the sub-level probability value set to obtain the corresponding fusion probability set for probe position discrimination, thus completing the probe position discrimination. This addresses the technical problem of difficulty in improving probe position discrimination performance with minimal human intervention.

[0213] Specifically, for a supervised pre-defined deep learning classification network or a pre-defined machine learning classifier, the area quantitative features corresponding to the first and second overlapping areas used for training, the features in the validation set, the sub-training set, and the sub-validation set are configured as area quantitative features and probe position discrimination labels corresponding to the gas.

[0214] Specifically, using the corresponding sub-preset probe position discrimination model for each preset classifier and the sub-validation sets of the first and second overlap areas used for training, the corresponding sub-classification probability value set for each preset classifier on the sub-validation set is determined. Similarly, using the corresponding preset probe position discrimination model for each preset classifier and the validation sets corresponding to the first and second overlap areas used for training, the corresponding classification probability value set for each preset classifier on the validation set is determined.

[0215] Specifically, constructing the preset classifier pool includes: obtaining a preset classifier with a set number ≥ 2; and constructing the preset classifier pool using the preset classifier with a set number ≥ 2. Simultaneously, constructing a preset model clinical guidance pool includes: obtaining preset model clinical guidance corresponding to the area under the receiver operating characteristic curve (ROC) and the coordinate axes, recall, and precision; and constructing the preset model clinical guidance pool using the preset model clinical guidance corresponding to the area under the ROC curve and the coordinate axes. The preset classifiers can be configured as deep learning-based classification networks corresponding to CNN, RNN, and GCN, random forest (RF), gradient boosting decision tree (GBDT), K-nearest neighbors (KNN), AdaBoost (Ada), and machine learning-based classifiers corresponding to linear discriminant analysis (LDA).

[0216] Specifically, before determining the corresponding preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the first overlap area and the second overlap area for training, and the corresponding sub-preset probe position discrimination model for each preset classifier corresponding to the sub-training set of the first overlap area and the second overlap area for training, the process includes: initializing the list of unprocessed fusion weights corresponding to each preset model clinical guidance in the preset model clinical guidance pool, initializing the list of probability value sets corresponding to the sub-grading probability value set and the grading probability value set, initializing the fusion probability set corresponding to each preset model clinical guidance, initializing the list of unprocessed fusion weights corresponding to each preset model clinical guidance in the preset model clinical guidance pool, initializing the list of fusion probability value sets corresponding to each preset model clinical guidance, and initializing the list of probability value sets corresponding to the sub-grading probability value set and the grading probability value set.

[0217] Specifically, the list of unprocessed fusion weights corresponding to each preset model clinical guidance in the initialized preset model clinical guidance pool includes: the unprocessed fusion weights corresponding to each preset model clinical guidance in the initialized preset model clinical guidance pool.

[0218] Specifically, the dataset is divided according to a first preset ratio corresponding to the number of folds in cross-validation, resulting in a training set and a validation set corresponding to a first and second overlapping area used for training. Simultaneously, the first and second overlapping areas in the training set are divided according to a second preset ratio corresponding to stratified sampling, resulting in sub-training sets and sub-validation sets. Those skilled in the art can configure the first and second preset ratios according to actual needs. For example, the first preset ratio can be configured as 4:1, and the second preset ratio as 9:1.

[0219] Specifically, according to the first preset ratio corresponding to the number of folds in cross-validation, the known probe insertion positions of the area quantitative feature corresponding to the dataset are divided to obtain a first overlapping area, a second overlapping area training set, and a validation set for training. According to the second preset ratio corresponding to stratified sampling, the first overlapping area and the second overlapping area used for training in the training set are divided to obtain a sub-training set and a sub-validation set.

[0220] Specifically, determining the corresponding preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the first overlapping area and the second overlapping area used for training, and the corresponding sub-preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the sub-training set of the first overlapping area and the second overlapping area used for training, includes: determining the corresponding preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the first overlapping area and the second overlapping area used for training, and the corresponding multiple sub-preset probe position discrimination models for each preset classifier in the preset classifier pool corresponding to the sub-training set of the first overlapping area and the second overlapping area used for training.

[0221] Specifically, determining the corresponding sub-preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the sub-training sets of the first and second overlapping areas used for training includes: using the area quantitative features corresponding to the sub-training sets of the first and second overlapping areas used for training, and using the area quantitative features to know the probe insertion position, training each preset classifier in the preset classifier pool to obtain the corresponding sub-preset probe position discrimination model.

[0222] Specifically, it also includes: dividing the set according to a first preset ratio to obtain a first overlapping area, a second overlapping area, and a validation set for training. It also includes: dividing the first overlapping area and the second overlapping area for training according to a second preset ratio to obtain a sub-training set and a sub-validation set; the preset classifier is configured as a deep learning-based hierarchical network or a machine learning-based classifier.

[0223] For example, specifically, the preset classifier can be configured as a deep learning-based classifier network (e.g., CNN, RNN, GCN, etc.) or a machine learning-based classifier (e.g., decision tree, Naive Bayes, support vector machine, multilayer perceptron, decision tree, random forest, gradient boosting tree, etc., one or more).

[0224] Specifically, determining the fusion weight list corresponding to the sub-level probability value set includes: using a preset weight function to perform weight processing on the fusion weight list to be processed corresponding to the sub-level probability value set to obtain the corresponding fusion weight list. The preset weight function is configured as one or more of a linear weight function, an exponential weight function, a power weight function, and a classifier ranking weight function.

[0225] Specifically, determining the list of fusion weights to be processed includes: calculating the area, recall, and precision under the subject operating characteristic curve corresponding to the clinical guidance of the preset model and the coordinate axis using the sub-prediction probability value set; and configuring the area, recall, and precision under the subject operating characteristic curve and the coordinate axis as the list of fusion weights to be processed.

[0226] Specifically, applying the linear weighting function to the list of fusion weights to be processed includes: summing the fusion weights in the list to be processed to obtain a first weight sum; dividing each fusion weight by the first weight sum to obtain a corresponding fusion weight list. Applying the power weighting function to the list of fusion weights to be processed includes: using a first adjustment parameter (configured to 2 or other possible values), exponentializing each fusion weight in the list to be processed (the first adjustment parameter is the power) to obtain processed fusion power weights; summing the processed fusion power weights to obtain a second weight sum; dividing each processed fusion power weight by the second weight sum to obtain a corresponding fusion weight list. The process of applying the exponential weight function to the list of fusion weights to be processed includes: multiplying the fusion weights in the list of fusion weights to be processed using a second adjustment parameter (configured to 5 or other possible values) to obtain adjusted fusion weights; exponentially processing the adjusted fusion weights (using the adjusted fusion weights as powers of the exponent) to obtain exponential weights; summing the exponential weights to obtain a third weight sum; and dividing the exponential weights by the third weight sum to obtain the corresponding fusion weight list. The process of applying the prediction model ranking weight function to the list of fusion weights to be processed includes: determining the ranking of the sub-predictive model corresponding to each predictor based on the area enclosed by the receiver operating characteristic curve and the coordinate axis, recall, and precision; summing the rankings corresponding to the clinical orientation of each preset model to obtain a fourth weight sum; and dividing the rankings corresponding to the clinical orientation of each preset model by the corresponding fourth weight sum to obtain the corresponding fusion weight list.

[0227] Specifically, determining the preset probe position discrimination model corresponding to each preset classifier in the preset classifier pool corresponding to the first overlapping area and the second overlapping area for training, and the sub-preset probe position discrimination model corresponding to each preset classifier in the preset classifier pool corresponding to the sub-training set of the first overlapping area and the second overlapping area for training, includes: determining the preset probe position discrimination model corresponding to each preset classifier in the preset classifier pool corresponding to the first overlapping area and the second overlapping area for training, and the multiple sub-preset probe position discrimination models corresponding to each preset classifier in the preset classifier pool corresponding to the sub-training set of the first overlapping area and the second overlapping area for training.

[0228] Specifically, determining the corresponding preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the first and second overlapping areas used for training includes: extracting features from the first and second overlapping areas used for training to obtain corresponding quantitative area features, and training each preset classifier in the preset classifier pool using the quantitative area features and their known probe insertion positions to obtain multiple corresponding preset probe position discrimination models.

[0229] Specifically, determining the corresponding sub-preset probe position discrimination model for each preset classifier in the preset classifier pool corresponding to the sub-training sets of the first and second overlapping areas used for training includes: using the area quantitative features corresponding to the sub-training sets of the first and second overlapping areas used for training, and using the area quantitative features and their known probe insertion positions to train each preset classifier in the preset classifier pool to obtain the corresponding sub-preset probe position discrimination model.

[0230] Specifically, the known probe insertion position is configured to be on one side or the other side of the tooth in the oral X-ray training image.

[0231] Specifically, the insertion depth of the dental probe into the gap between a predetermined side of the tooth and its adjacent periodontal tissue is configured to be the tooth boundary line within the periodontal tissue. Simultaneously, a predetermined probe depth estimation model can be constructed, and the insertion depth of the dental probe into the gap between a predetermined side of the tooth and its adjacent periodontal tissue can be estimated based on this model.

[0232] Specifically, it also includes: extracting the overlapping area between a preset side of the tooth and the periodontal tissue from the oral X-ray image to be analyzed; and estimating the probe insertion depth of the tooth probe into the gap between the preset side of the tooth and its adjacent periodontal tissue based on a preset probe depth estimation model, using the overlapping area and the corresponding tooth boundary line within the periodontal tissue.

[0233] Specifically, constructing the preset probe depth estimation model includes: extracting the overlapping area formed by periodontal tissues adjacent to a preset side of the tooth used for training and the corresponding tooth boundary line within the periodontal tissue from a preset number of oral X-ray training images; training a preset classifier using the overlapping area formed by the periodontal tissues adjacent to the preset side of the tooth used for training, the corresponding tooth boundary line within the periodontal tissue, and the corresponding known probe depth, to construct the preset probe depth estimation model. More specifically, the preset classifier can be configured as a deep learning-based classification network corresponding to CNN, RNN, GCN, random forest (RF), gradient boosting decision tree (GBDT), K-nearest neighbors (KNN), AdaBoost (Ada), and a machine learning-based classifier corresponding to linear discriminant analysis (LDA).

[0234] Specifically, the step of extracting the tooth boundary line within the periodontal tissue corresponding to the preset side of the tooth where the tooth probe is inserted from the oral X-ray image to be analyzed includes: extracting the tooth mask edge coordinates of the tooth mask image corresponding to the tooth and the periodontal tissue mask image corresponding to the preset side of the tooth from the oral X-ray image to be analyzed; and determining the tooth boundary line within the periodontal tissue on the periodontal tissue mask image based on the tooth mask edge coordinates.

[0235] Specifically, determining the tooth mask image includes: segmenting the teeth of the oral X-ray image to be analyzed using a deep learning-based tooth segmentation model to obtain a corresponding tooth mask image; and / or determining the tooth mask edge coordinates includes: detecting the tooth mask lines corresponding to the tooth mask image; and extracting the tooth mask edge coordinates of the tooth mask lines.

[0236] Specifically, a deep learning-based periodontal tissue segmentation model is used to segment the periodontal tissue in the oral X-ray image to be analyzed, obtaining a corresponding periodontal tissue mask image. More specifically, the step of extracting the overlapping area formed by the periodontal tissue and the adjacent teeth on both sides of the periodontal tissue from the oral X-ray image to be analyzed includes: segmenting the periodontal tissue and teeth in the oral X-ray image to be analyzed, obtaining a periodontal tissue mask image and a tooth mask image; determining the overlapping area based on the periodontal tissue mask image and the tooth mask image; or, segmenting the periodontal tissue and teeth in the oral X-ray image to be analyzed, obtaining a periodontal tissue and tooth overlapping mask image; determining the overlapping area based on the periodontal tissue and tooth overlapping mask image.

[0237] Specifically, determining the overlapping area based on the periodontal tissue mask image and the tooth mask image includes: extracting a first mask region corresponding to the periodontal tissue mask in the periodontal tissue mask image and a second mask region corresponding to the tooth mask in the tooth mask image; and determining the common area of ​​the first mask region and the second mask region as the overlapping area.

[0238] Similarly, the step of extracting the overlapping area for training from a preset number of oral X-ray training images includes: extracting the third mask region corresponding to the periodontal tissue mask in the periodontal tissue mask training image corresponding to each oral X-ray training image in the preset number of oral X-ray training images and the fourth mask region corresponding to the tooth mask in the tooth mask training image; and determining the common area of ​​the third mask region and the fourth mask region as the overlapping area for training.

[0239] Specifically, determining the overlapping area based on the overlapping mask image of periodontal tissue and teeth includes: calculating the area corresponding to the overlapping mask value of periodontal tissue and teeth in the overlapping mask image of periodontal tissue and teeth, and determining the overlapping area.

[0240] Specifically, determining the periodontal tissue mask training image and the tooth mask training image corresponding to each of the preset number of oral X-ray training images includes: using segmentation and annotation software to delineate the periodontal tissue and teeth in each of the preset number of oral X-ray training images, respectively, to obtain the corresponding periodontal tissue mask training image and the tooth mask training image. The segmentation and annotation software can be configured as CasiaLabeler, Labelme, or other possible annotation software.

[0241] Specifically, a pre-defined segmentation model based on deep learning is used to segment the periodontal tissue and teeth in the oral X-ray image to be analyzed, resulting in a periodontal tissue mask image and a tooth mask image. The periodontal tissue mask in the periodontal tissue mask image can be configured with a first mask value, and the tooth mask in the tooth mask image can be configured with a second mask value that is different from the first mask value. Those skilled in the art can configure the first mask value and the second mask value according to actual needs. For example, the first mask value can be configured to 1, and the second mask value can be configured to 2.

[0242] Specifically, the step of extracting the overlapping area formed by periodontal tissue and adjacent teeth on both sides of the periodontal tissue from the oral X-ray image to be analyzed further includes: constructing a preset segmentation model based on deep learning, and using the preset segmentation model to segment the periodontal tissue and teeth in the oral X-ray image to be analyzed to obtain a periodontal tissue mask image and a tooth mask image.

[0243] Specifically, constructing the deep learning-based preset segmentation model includes: training the preset deep learning segmentation model using a preset number of oral X-ray training images and their corresponding periodontal tissue and tooth mixed mask training images, thereby constructing the deep learning-based preset segmentation model. This addresses at least one of the technical problems arising from the current requirement for manual segmentation of periodontal tissue and teeth, which results in a huge workload for doctors and prevents further technical analysis of kidney disease early warning related to periodontal plaque.

[0244] Specifically, unlike existing technologies, a mixed mask training image of periodontal tissue and teeth is constructed. During the training of the preset deep learning segmentation network, the preset deep learning segmentation network pays more attention to the anatomical relationship of the periodontal tissue and teeth, forming a collaborative learning and collaborative response to the periodontal tissue and teeth. This enables effective anatomical segmentation of the periodontal tissue and teeth, and creates anatomical constraints on the segmentation of the periodontal tissue and teeth.

[0245] Specifically, the construction of the periodontal tissue and tooth mixed mask training image includes: mapping the first mask value corresponding to the periodontal tissue mask training image of each oral X-ray training image in the preset number of oral X-ray training images to the tooth mask training image according to its corresponding first position coordinates, thereby constructing the periodontal tissue and tooth mixed mask training image; or, mapping the second mask value corresponding to the tooth mask training image of each oral X-ray training image in the preset number of oral X-ray training images to the corresponding periodontal tissue mask training image according to its corresponding second position coordinates, thereby constructing the periodontal tissue and tooth mixed mask training image.

[0246] Specifically, the step of extracting the overlapping area formed by periodontal tissue and adjacent teeth on both sides of the periodontal tissue from the oral X-ray image to be analyzed further includes: constructing a preset segmentation model based on deep learning, using the preset segmentation model to segment the periodontal tissue and teeth in the oral X-ray image to be analyzed, and performing periodontal tissue and tooth segmentation on the oral X-ray image to be analyzed to obtain a periodontal tissue and tooth overlap mask image.

[0247] Specifically, the construction of the periodontal tissue and tooth mixed mask training image includes: mapping the first mask value corresponding to the periodontal tissue mask training image of each oral X-ray training image in the preset number of oral X-ray training images to the tooth mask training image according to its corresponding first position coordinates, to obtain the first mask value and the second mask value plus the corresponding overlapping mask value, so as to construct the periodontal tissue and tooth overlapping mask training image; or, mapping the second mask value corresponding to the tooth mask training image of each oral X-ray training image in the preset number of oral X-ray training images to the corresponding periodontal tissue mask training image according to its corresponding second position coordinates, to obtain the second mask value and the first mask value plus the corresponding overlapping mask value, so as to construct the periodontal tissue and tooth overlapping mask training image.

[0248] Specifically, the periodontal tissue and tooth mixed mask training image includes: a first mask value corresponding to the periodontal tissue mask training image, a second mask value corresponding to the tooth mask training image, and an overlapping mask value corresponding to the sum of the first and second mask values. Unlike the periodontal tissue and tooth mixed mask training image, the overlapping mask in the periodontal tissue and tooth overlapping mask training image represents both the first mask value corresponding to the periodontal tissue mask and the second mask value corresponding to the tooth mask.

[0249] Specifically, training the preset deep learning segmentation model to construct the preset deep learning-based segmentation model includes: during the training of the preset deep learning segmentation network using the mixed periodontal tissue and tooth mask training image (or the overlapping periodontal tissue and tooth mask training image) and its corresponding preset number of oral X-ray training images, obtaining corresponding periodontal tissue mask images and tooth mask images (or corresponding overlapping periodontal tissue and tooth mask images); using a preset periodontal tissue and tooth overlapping edge encoder, encoding the periodontal tissue and tooth edge mask images of the periodontal tissue mask images and tooth mask images (or the overlapping periodontal tissue and tooth mask images), and the edge mask training images of the mixed periodontal tissue and tooth mask images (or the overlapping periodontal tissue and tooth mask images), respectively, to obtain a first edge encoding feature map and a second edge encoding feature map; calculating the loss value between the first edge encoding feature map and the second edge encoding feature map to construct the preset deep learning-based segmentation model.

[0250] Specifically, those skilled in the art can configure the loss function corresponding to the loss value according to actual needs. For example, the loss function can be configured as one or more of the following loss functions: Cross-Entropy Loss, Dice Loss, Tversky Loss, and Focal Loss.

[0251] Specifically, determining the edge mask training image includes: performing an erosion operation on the periodontal tissue and tooth mixed mask training image (or the periodontal tissue and tooth overlapping mask training image) to obtain a periodontal tissue and tooth mixed mask erosion image (or the periodontal tissue and tooth overlapping mask training erosion image); performing a pixel-level subtraction operation on the periodontal tissue and tooth mixed mask training image (or the periodontal tissue and tooth overlapping mask training image) and the corresponding periodontal tissue and tooth mixed mask training erosion image (or the periodontal tissue and tooth overlapping mask training erosion image) to determine the edge mask training image.

[0252] Specifically, the construction of the preset periodontal tissue and tooth overlap edge encoder includes: constructing the periodontal tissue and tooth overlap edge encoder by using the set number of edge mask training images and their corresponding edge training images with gray values.

[0253] Specifically, the step of constructing a periodontal tissue and tooth overlap edge encoder by using the set number of edge mask training images and their corresponding edge training images with gray values ​​includes: extracting edge training images with gray values ​​corresponding to the oral X-ray training images; and using the edge training images and their corresponding edge mask training images to train a preset downsampling convolutional network to construct a periodontal tissue and tooth overlap edge encoder.

[0254] Specifically, the step of extracting the edge training image with grayscale value corresponding to the oral X-ray training image includes: performing a pixel-level multiplication operation on the edge mask training image corresponding to the set number and the oral X-ray training image to obtain the edge training image with grayscale value.

[0255] Specifically, unlike existing technologies, the preset periodontal tissue and tooth overlap edge encoder enables the edges corresponding to the periodontal tissue and tooth overlap within the liver cancer tumor to participate in the training process of the preset deep learning segmentation network, thereby achieving controllability of the edges corresponding to the periodontal tissue and tooth overlap, which is more conducive to the segmentation of the edges of the periodontal tissue and tooth overlap.

[0256] Specifically, during the construction of the preset periodontal tissue and tooth overlap edge encoder, the loss value between the periodontal tissue mask image and the tooth mask image (or the periodontal tissue and tooth overlap mask image) and the edge mask training image of the periodontal tissue and tooth hybrid mask (or the periodontal tissue and tooth overlap mask training image) is calculated. Similarly, those skilled in the art can configure the loss function corresponding to the loss value according to actual needs. For example, the loss function can be configured as one or more of the following loss functions: Cross-Entropy Loss, Dice Loss, Tversky Loss, and Focal Loss.

[0257] Specifically, the preset downsampling convolutional network can be configured as one or more of the downsampling convolutional networks corresponding to ResNet, EfficientNet, LeNet, AlexNet, YOLOv10, SRFD / DRFD modules, HWD modules, MobileNet, ShuffleNet, etc. Those skilled in the art can configure the kernel size, stride, pooling, and other conventional layers of the aforementioned preset downsampling convolutional network according to actual needs.

[0258] Specifically, the preset deep learning segmentation network is configured as one or more deep learning segmentation networks such as Fully Convolutional Network (FCN), U-Net, DeepLab, and Pyramid Spatial Pooling Network (PSPNet).

[0259] The method also includes: estimating the probe insertion depth corresponding to the oral X-ray image to be analyzed in a patient with severe periodontitis using the method described above; if the probe insertion depth is greater than a preset probe insertion depth and the alveolar bone resorption on a preset side of the tooth corresponding to the oral X-ray image is greater than a preset length ratio of the root length, then periodontal plaque is collected using the tooth probe; and based on the periodontal plaque, a kidney disease warning associated with the periodontal plaque is generated. The preset probe insertion depth can be configured to 5 mm or other possible values. The preset length ratio can be configured to 1 / 2 or other possible values.

[0260] Specifically, before collecting periodontal plaque using the dental probe, the procedure includes: after the patient to be collected has fasted for a first preset time, the patient holds a first preset volume of water in their mouth and rinses their mouth for a second preset time; after repeating the holding of the preset volume of water in the mouth and rinsing the mouth for a preset number of times corresponding to the second preset time, periodontal plaque is collected using a clinical probe and placed in an Eppendorf tube containing a second preset volume of sodium thiosulfate preservation solution, and the Eppendorf tube is stored at a preset temperature.

[0261] Specifically, the kidney disease can be configured as immunoglobulin A nephropathy. Additionally, if the probe insertion depth is greater than a preset probe insertion depth but less than the preset probe insertion depth, the probe insertion depth on the anterior or posterior side of the tooth surface is estimated.

[0262] Specifically, estimating the probe insertion depth on the anterior or posterior side of the tooth surface includes: extracting the tooth boundary line within the periodontal tissue corresponding to the insertion of the probe on the anterior or posterior side of the tooth surface from the oral X-ray image to be analyzed; and estimating the probe insertion depth between the preset side of the tooth and its adjacent periodontal tissue based on the tooth boundary line within the periodontal tissue. In this case, there are no adjacent teeth on the anterior or posterior side of the tooth surface in the oral X-ray image to be analyzed; it is only necessary to configure the anterior or posterior side of the tooth surface as the preset side of the tooth.

[0263] Specifically, the overlapping area formed by the front or back of the tooth surface with the periodontal tissue is extracted from the oral X-ray image to be analyzed; based on a preset probe depth estimation model, the probe insertion depth of the tooth probe into the gap between the front or back of the tooth surface and the adjacent periodontal tissue is estimated by using the overlapping area and the corresponding tooth boundary line in the periodontal tissue.

[0264] Specifically, constructing the preset probe depth estimation model includes: extracting the overlapping area formed by adjacent periodontal tissues on the front or back of the tooth surface for training from a preset number of oral X-ray training images, and the corresponding tooth boundary line within the periodontal tissue; training a preset classifier using the overlapping area formed by adjacent periodontal tissues on a preset side of the tooth for training, the corresponding tooth boundary line within the periodontal tissue, and the corresponding known probe depth, to construct the preset probe depth estimation model.

[0265] Specifically, before collecting periodontal plaque using the dental probe, the procedure includes: after the patient has fasted for a first preset time of 30 minutes, the patient holds 150ml of water in their mouth and rinses their mouth for a second preset time; after repeating the process 2-3 times (corresponding to the preset number of times of holding the preset volume of water in the mouth and rinsing the mouth thoroughly for 10 seconds within the second preset time), periodontal plaque is collected using the clinical probe and placed in an Eppendorf tube containing 1ml of sodium thiosulfate preservation solution, labeled, and stored in a refrigerator at a preset temperature of -80 degrees Celsius. Periodontal plaque was collected from 100 patients using the above method.

[0266] Specifically, the method of providing early warning of kidney disease associated with periodontal plaque based on the periodontal plaque includes: providing early warning of kidney disease recurrence and / or early warning of kidney disease risk based on biomarkers corresponding to periodontal plaque in the disease group with kidney disease (IgA nephropathy) combined with chronic periodontitis (severe chronic periodontitis) and the control group with chronic periodontitis (one or more of severe chronic periodontitis, periodontal disease, and gingival crevicular fluid inflammation) without kidney disease (IgA nephropathy).

[0267] Specifically, the method of using biomarkers corresponding to periodontal plaque in the disease group of IgA nephropathy combined with chronic periodontitis (severe chronic periodontitis) and the control group of chronic periodontitis (one or more of severe chronic periodontitis, periodontal disease, and gingival crevicular fluid inflammation) without IgA nephropathy to perform relapse warning and / or risk warning for kidney disease includes: using biomarkers corresponding to periodontal plaque in the disease group of IgA nephropathy combined with chronic periodontitis (severe chronic periodontitis) and the control group of chronic periodontitis (one or more of severe chronic periodontitis, periodontal disease, and gingival crevicular fluid inflammation) without IgA nephropathy to train a preset classifier to construct a relapse warning model and / or risk warning model for kidney disease; and using biomarkers corresponding to the oral X-ray to be analyzed based on the relapse warning model and / or risk warning model for kidney disease to perform relapse warning and / or risk warning for kidney disease.

[0268] Specifically, before acquiring the oral X-ray image to be analyzed, the method includes: acquiring multiple positioning trigger signals corresponding to the positioning and mating component of the foot pedal connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame; and positioning the foot pedal and the bed frame at preset positioning positions corresponding to the preset positioning signals based on the multiple positioning trigger signals and preset positioning signals. This addresses at least one of the technical problems currently encountered with foot pedals or digital X-ray machines (bedside X-ray machines), such as the difficulty in quickly releasing the foot pedal and bed frame from positioning and quickly assembling and positioning them during handling or assembly.

[0269] Specifically, a digital X-ray machine (bedside X-ray machine) is a portable digital X-ray device designed specifically for critically ill or immobile patients, allowing for imaging examinations of the chest, limbs, and other areas at the bedside. Its compact size and flexible power supply avoid the risks associated with patient transport, making it particularly suitable for emergency scenarios such as ICUs and emergency rooms. The foot pedal of the digital X-ray machine is a manual control device used to trigger X-ray exposure. It allows for quick switching between fluoroscopy and spot imaging modes to support real-time intraoperative imaging, enabling the acquisition of corresponding oral X-ray images for analysis. Simultaneously, the foot pedal can also control X-ray exposure for acquiring corresponding oral X-ray images for analysis.

[0270] Specifically, the bed frame includes: a first bed frame and / or a second bed frame disposed on the opposite side of the first bed frame; footrests are respectively disposed at one end of the first bed frame and one end of the second bed frame. The first bed frame is disposed parallel to the second bed frame. That is, the first bed frame and the second bed frame can be configured to be parallel or, depending on actual needs, non-parallel, but one end of the first bed frame and one end of the second bed frame are respectively connected to the two ends of the footrests.

[0271] Specifically, based on the multiple positioning trigger signals and the preset positioning signal, the footrest and the bed frame are positioned at the preset positioning positions corresponding to the preset positioning signals, adjusting the relative position between the bed frame and the footrest to be positioned. This is to prevent the footrest and bed frame from being easily disengaged during transport or easily assembled during assembly. More specifically, the positioning body corresponding to the preset positioning position is configured as a groove, and the positioning mating component or the positioning mating part corresponding to the positioning mating component is configured as a protrusion matching the groove. When force is applied to the positioning mating component, the applied force acts on the positioning mating component, and the positioning mating component drives the footrest to move along the position of the positioning body corresponding to the preset positioning position. When the positioning mating component moves to the position of the positioning body, the positioning body connects with the positioning mating component and emits a "click" sound, indicating that the relative position between the bed frame and the footrest has been adjusted to the preset relative position.

[0272] Specifically, before acquiring the multiple positioning trigger signals corresponding to the positioning mating component connected to the bed frame during its movement towards the multiple preset positioning positions corresponding to the bed frame, the process includes: applying a first force to the positioning mating component during its movement towards the multiple preset positioning positions corresponding to the bed frame and applying a second force to the positioning mating component to generate the multiple positioning trigger signals; under the action of the first force and the second force, determining the multiple positioning trigger signals corresponding to the positioning mating component connected to the bed frame during its movement towards the multiple preset positioning positions corresponding to the bed frame.

[0273] Specifically, when a force (first force) is applied to the positioning and fitting component, the force along the positioning body corresponding to the preset position to be positioned causes the foot pedal to move along the position of the positioning body corresponding to the preset position to be positioned; the force (second force) acting on the positioning and fitting component generates the plurality of positioning trigger signals.

[0274] Specifically, positioning the footrest and the bed frame at a preset positioning position corresponding to a preset positioning signal includes: determining the cumulative trigger count corresponding to the plurality of positioning trigger signals; and positioning the footrest and the bed frame at the preset positioning position corresponding to the preset positioning signal based on the cumulative trigger count and the preset cumulative trigger count corresponding to the preset positioning signal. Specifically, the positioning body corresponding to the preset position to be positioned is configured as a plurality of positioning bodies corresponding to different preset positions to be positioned; the positioning mating component can sequentially contact the plurality of positioning bodies corresponding to the different preset positions to be positioned to generate the plurality of positioning trigger signals.

[0275] Specifically, the step of positioning the footrest and the bed frame at the preset positioning position corresponding to the preset positioning signal based on the cumulative trigger count and the preset cumulative trigger count corresponding to the preset positioning signal includes: if the cumulative trigger count is equal to the preset cumulative trigger count, then the footrest and the bed frame are positioned at the preset positioning position corresponding to the preset positioning signal. For example, specifically, the preset cumulative trigger count corresponding to the preset positioning signal is configured to be 2. The positioning mating component can sequentially contact the second positioning body corresponding to the different preset positions to be positioned to generate 2 positioning trigger signals. At this time, the cumulative trigger count corresponding to the 2 positioning trigger signals is configured to be 2, and the footrest and the bed frame are positioned to move to the preset positioning position corresponding to the preset positioning signal.

[0276] Specifically, determining multiple positioning trigger signals corresponding to the positioning and mating component of the footrest connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame includes: acquiring audio signals corresponding to the positioning and mating component of the footrest connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame; and based on the audio signals, determining multiple positioning trigger signals corresponding to the positioning and mating component of the footrest connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame. Specifically, an audio acquisition unit (e.g., a recorder or microphone) can be used to acquire the audio signals corresponding to the positioning and mating component of the footrest connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame.

[0277] Specifically, the method of determining multiple positioning trigger signals corresponding to the positioning and mating component of the footrest connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame, based on the audio signal, includes: acquiring a preset spectrum corresponding to the contact and mating process between the positioning and mating component and the positioning body corresponding to the multiple preset positioning positions; and determining multiple positioning trigger signals corresponding to the preset spectrum based on the audio signal and the preset spectrum during the movement of the positioning and mating component of the footrest connected to the bed frame towards the multiple preset positioning positions corresponding to the bed frame.

[0278] Specifically, a Fourier transform is performed on the audio signal corresponding to the positioning and mating component of the footrest connected to the bed frame during its movement towards multiple preset positioning positions corresponding to the bed frame, to obtain an audio spectrum; a preset spectrum corresponding to the moment when the positioning body connects with the positioning and mating component (emitting a "click" sound) is extracted from the audio spectrum. A preset audio signal corresponding to the moment when the positioning body connects with the positioning and mating component (emitting a "click" sound) is extracted from the audio signal corresponding to the movement of the positioning and mating component of the footrest connected to the bed frame towards multiple preset positioning positions corresponding to the bed frame; a Fourier transform is performed on the preset audio signal to obtain a preset spectrum.

[0279] Specifically, based on the plurality of positioning trigger signals and the preset positioning signal, the footrest and the bed frame are positioned at the preset positioning positions corresponding to the preset positioning signals. This also includes using a limiting mechanism to limit and fix the preset position to be positioned after positioning. When a second force generating the plurality of positioning trigger signals is applied to the positioning mating component, the limiting mechanism is controlled to disengage from the bed frame; when the second force generating the plurality of positioning trigger signals is removed from the positioning mating component, the limiting mechanism is controlled to limit the bed frame.

[0280] For example, multiple positioning bodies, including: a first positioning body and a second positioning body; when disengaging from the first or second positioning body corresponding to the multiple positioning bodies, firstly, a force is applied to the outer handle surface of the handle corresponding to the drive mechanism (a force along the direction of movement of the positioning body corresponding to the preset positioning position), thereby releasing the connection of the first or second positioning body corresponding to the multiple positioning bodies (the limiting head of the limiting mechanism disengages from the limiting groove set in the first or second positioning body corresponding to the multiple positioning bodies), and reconnects with the positioning and mating components of the foot pedal according to the preset positioning position of the first or second positioning body corresponding to the multiple positioning bodies.

[0281] Specifically, it includes: a bed frame and a footrest connected to the bed frame; the bed frame is provided with a positioning body corresponding to a preset positioning position; the positioning body corresponding to the preset positioning position is connected to and disconnected from the positioning engagement component of the footrest; the positioning body and the positioning engagement component are used to adjust the relative position between the bed frame and the footrest. This solves at least one of the technical problems currently present, such as the difficulty in quickly releasing positioning and quickly assembling positioning during handling or assembly.

[0282] Specifically, when a force is applied to the positioning and fitting component, the applied force acts on the positioning and fitting component, and the positioning and fitting component drives the foot pedal to move along the position of the positioning body corresponding to the preset positioning position. When the positioning and fitting component moves to the position of the positioning body, the positioning body connects with the positioning and fitting component and makes a "click" sound, thereby indicating that the relative position between the bed frame and the foot pedal has been adjusted to the preset relative position.

[0283] Specifically, the positioning and engaging assembly includes: a sliding body disposed on the bed frame and a positioning and engaging component disposed on the sliding body; the sliding body is slidably connected to the bed frame and is used to drive the positioning and engaging component of the sliding body to engage with the positioning body. The sliding body includes: a telescopic shaft, a compression spring connected to the telescopic shaft, and a positioning and engaging component connected to the compression spring. When a force is applied to the positioning and engaging assembly, the force along the positioning body corresponding to the preset positioning position drives the foot pedal to move along the position of the positioning body corresponding to the preset positioning position. The force acting on the telescopic shaft of the positioning and engaging assembly causes the telescopic shaft of the positioning and engaging assembly to retract. When the telescopic shaft retracts, it compresses the compression spring connected to the telescopic shaft. The compressed compression spring acts on the positioning and engaging component. When the telescopic shaft of the positioning and engaging assembly moves to the position of the positioning body, the positioning body connects with the telescopic shaft of the positioning and engaging assembly and emits a "click" sound, thereby indicating that the relative position between the bed frame and the foot pedal has been adjusted to the preset relative position.

[0284] Specifically, the sliding body further includes: a bushing fitted onto one end of the telescopic shaft; a compression spring connected to one end of the telescopic shaft and located within the bushing; and a positioning fitting component connected to the compression spring and capable of ejecting the bushing. When force is applied to the positioning fitting component, the force along the positioning body corresponding to the preset positioning position drives the foot pedal to move along the position of the positioning body corresponding to the preset positioning position. The force acting on the telescopic shaft of the positioning fitting component causes the telescopic shaft of the positioning fitting component to retract. When the telescopic shaft retracts, it compresses the compression spring connected to the telescopic shaft. The compressed compression spring acts on the positioning fitting component connected to the compression spring and capable of ejecting the bushing, ejecting the bushing. When the telescopic shaft of the positioning fitting component moves to the position of the positioning body, the positioning body connects with the telescopic shaft of the positioning fitting component and emits a "click" sound, thereby indicating that the relative position between the bed frame and the foot pedal has been adjusted to the preset relative position.

[0285] Specifically, the positioning and mating assembly further includes: a drive mechanism connected to the sliding body; the drive mechanism is used to drive the sliding body to slide on the bed frame. The force applied by the drive mechanism to drive the sliding body to slide on the bed frame, when applied to the positioning and mating assembly, causes the foot pedal to move along the position of the positioning body corresponding to the preset positioning position. The force acting on the telescopic shaft of the positioning and mating assembly causes the telescopic shaft to retract. When the telescopic shaft retracts, it compresses the spring connected to the telescopic shaft. The compressed spring acts on the positioning and mating component connected to the spring and capable of ejecting the bushing, causing it to eject the bushing. When the telescopic shaft of the positioning and mating assembly moves to the position of the positioning body, the positioning body connects to the telescopic shaft of the positioning and mating assembly, and a "click" sound is emitted, indicating that the relative position between the bed frame and the foot pedal has been adjusted to the preset relative position. When the force at the other end of the telescopic shaft of the sliding body is withdrawn, the telescopic shaft changes from the contracted state to the natural state, and the compression spring is no longer applied, and the positioning fitting retracts into the bushing.

[0286] Specifically, the positioning and mating assembly further includes: a housing disposed on the outer side of the sliding body, with the other end of the telescopic shaft of the sliding body passing through the housing and connected to a drive mechanism. The drive mechanism connected to the other end of the telescopic shaft of the sliding body through the housing is configured as a handle. When the handle corresponding to the drive mechanism drives the sliding body to slide on the bed frame, the force applied to the positioning and mating assembly causes the foot pedal to move along the position of the positioning body corresponding to the preset positioning position. The force acting on the other end of the telescopic shaft of the sliding body causes the telescopic shaft of the positioning and mating assembly to retract. When the telescopic shaft retracts, it compresses the spring connected to the telescopic shaft. The compressed spring acts on the positioning and mating component connected to the spring and capable of ejecting the bushing, ejecting the bushing. When the telescopic shaft of the positioning and mating assembly moves to the position of the positioning body, the positioning body connects to the telescopic shaft of the positioning and mating assembly, and a "click" sound is emitted, indicating that the relative position between the bed frame and the foot pedal has been adjusted to the preset relative position. When the force at the other end of the telescopic shaft of the sliding body is withdrawn, the telescopic shaft changes from the contracted state to the natural state, and the compression spring is no longer applied, and the positioning fitting retracts into the bushing.

[0287] Specifically, the positioning mechanism further includes a first slider disposed between the bed frame and the main body housing. A second slider is disposed on the opposite side of the bed frame, where the first slider is disposed between the bed frame and the main body housing. The first slider is disposed on one side of the bed frame or the sliding body of the bed frame, and the second slider is disposed on the other side of the bed frame. The first and second sliders are used to clamp the bed frame vertically, thereby fixing the sliding body of the entire positioning mechanism vertically. The first and second sliders are configured as self-lubricating sliders with low hardness (less than a preset hardness), such as plastic blocks, to facilitate movement and prevent scratching the bed frame.

[0288] Specifically, the positioning body corresponding to the preset positioning position is configured as a groove, and the positioning mating component or the positioning mating part corresponding to the positioning mating component is configured as a protrusion matching the groove. It also includes: the limiting mechanism disposed on the bed frame; the limiting mechanism includes: a limiting head and a rotatable limiting base connected to the limiting head. A first slider groove is provided on one side of the bed frame, and the first slider slides within the first slider groove along the sliding direction of the positioning mating component. The limiting mechanism is respectively provided on both sides of the bed frame.

[0289] Specifically, the positioning body corresponding to the preset position to be positioned is configured as multiple positioning bodies corresponding to different preset positions to be positioned. For example, multiple positioning bodies include: a first positioning body and a second positioning body; when detached from the first positioning body or the second positioning body, firstly, a force is applied to the outer handle surface of the handle corresponding to the drive mechanism (the force along the direction of movement of the positioning body corresponding to the preset position to be positioned), and the limiting head of the limiting mechanism is released from the limiting groove of the first positioning body and the second positioning body corresponding to the bed frame, and reconnected to the positioning and mating component of the foot pedal according to the preset position to be positioned of the first positioning body or the second positioning body.

[0290] Specifically, when force is applied to the handle corresponding to the drive mechanism on the outer handle surface, the limiting head of the limiting mechanism disengages from the limiting groove inserted into the bed frame. Simultaneously, when force is applied to the handle corresponding to the drive mechanism on the outer handle surface, the force along the positioning body corresponding to the preset positioning position drives the foot pedal to move along the position of the positioning body corresponding to the preset positioning position. The force acting on the other end of the telescopic shaft of the sliding body (the force applied to the outer handle surface acts on the other end of the telescopic shaft of the sliding body) causes the telescopic shaft to retract, and the compression spring connected to the telescopic shaft to retract under force, forcing the positioning fitting connected to the compression spring and capable of popping out of the bushing to pop out of the bushing so that the positioning fitting can connect to the positioning body corresponding to the preset positioning position (at this time, a "click" sound is emitted). At this time, no force is applied to the handle corresponding to the drive mechanism on the outer handle surface, and the direction of the limiting head is adjusted by using the rotatable limiting base connected to the limiting head. The limiting head of the limiting mechanism is inserted into the limiting groove provided on the bed frame to realize the positioning limitation of the bed frame and the foot pedal, thereby realizing the adjustment of the relative position between the bed frame and the foot pedal.

[0291] Specifically, the limiting head includes: a limiting head body and multiple limiting holes (at least one limiting hole) disposed on the limiting head body. Multiple limiting screws are used to fix the limiting mechanism to the limiting grooves provided on the bed frame, passing through the multiple limiting holes respectively. When the limiting mechanism is fixed to the bed frame using the multiple limiting screws passing through the multiple limiting holes respectively, to release the connection between the limiting mechanism and the first or second positioning body, it is first necessary to remove the fixing connection between the multiple limiting screws and the limiting grooves provided on the bed frame. A second slider is provided on the other side of the first slider groove provided on the bed frame, and the second slider slides within the second slider groove along the sliding direction of the positioning and mating assembly.

[0292] Specifically, the second slider includes: a second slider body and a positioning body connector disposed on the second slider body; the positioning body connector includes: a positioning body connecting body and a plurality of connecting holes (at least one connecting hole) disposed on the positioning body. A plurality of limiting screws are used to pass through the plurality of limiting holes and connect to the limiting grooves provided on the bed frame to fix one side of the bed frame in a fixed connection.

[0293] Specifically, by pressing down on the first handle, a force is applied to the housing of the first handle in the direction of movement of the positioning body corresponding to the preset positioning position. The housing of the first handle is pressed against one side surface of the bed frame, which releases the connection between the limiting mechanism and the positioning body. The limiting head of the limiting mechanism disengages from the limiting groove corresponding to the bed frame. While pressing down on the first handle, the inside of the palm presses against one end of the telescopic shaft. The telescopic shaft retracts, and the compression spring connected to the telescopic shaft contracts under force, forcing the positioning fitting connected to the compression spring and capable of popping out of the bushing to pop out of the bushing, so that the positioning fitting corresponding to the preset positioning position of the first positioning body... The positioning body is connected (a "click" sound is heard at this time); if the first positioning body corresponding to the preset positioning position is the specified preset positioning position, the force applied to the housing of the first handle along the direction of movement of the positioning body corresponding to the preset positioning position and the force between the inside of the palm and one end of the telescopic shaft are released. At this time, the telescopic shaft returns from retracted to its natural state, and the compression spring connected to the telescopic shaft also returns from retracted to its natural state. The positioning fitting part connected to the compression spring and capable of popping out the bushing retracts to the inside of the bushing. The limiting head of the limiting mechanism connecting the limiting mechanism and the positioning body is inserted into the limiting groove corresponding to the bed frame to achieve rapid assembly and positioning. If the first positioning body corresponding to the preset positioning position is not the specified preset positioning position, then only the force between the inside of the palm and one end of the telescopic shaft is released. The positioning fitting part connected to the compression spring and capable of popping out the bushing retracts to the inside of the bushing. Under the force applied to the housing of the first handle along the direction of movement of the positioning body corresponding to the preset positioning position, it continues to move in the direction of movement of the positioning body corresponding to the next preset positioning position. After the positioning fitting part disengages from the first positioning body, the force between the inside of the palm and one end of the telescopic shaft continues until the preset positioning position is reached, achieving rapid assembly and positioning. If rapid release of positioning is achieved, only the force is applied to the housing of the first handle (no force is applied to one end of the telescopic shaft), and the limiting head of the limiting mechanism connecting the limiting mechanism and the positioning body disengages from the limiting groove corresponding to the bed frame.

[0294] Specifically, a first handle and a second handle are respectively provided at one end of the first bed frame and one side of the second bed frame; the first handle and the second handle are respectively provided with shaft holes corresponding to the telescopic shafts of the two sliding bodies, and one end of the telescopic shaft of the two sliding bodies passes through the corresponding shaft holes to form a protrusion mechanism on the outer handle surface of the first handle and the second handle, respectively. By applying force to the protrusion mechanism, the telescopic shaft retracts, the compression spring connected to the telescopic shaft retracts under force, and forces the positioning fitting member connected to the compression spring and capable of popping out of the bushing to pop out of the bushing, so that the positioning fitting member can connect to the positioning body corresponding to the preset positioning position.

[0295] Specifically, the first bed frame and the second bed frame are respectively equipped with the positioning and mating components. Each positioning and mating component includes a housing corresponding to the first bed frame and a housing corresponding to the second bed frame, disposed on the outer side of the sliding body. The first handle and the second handle are respectively connected to one side of the housing corresponding to the first bed frame and the housing corresponding to the second bed frame, such as by welding. The housing corresponding to the first bed frame and the housing corresponding to the second bed frame are respectively slidably connected to the other side of the first bed frame and the other side of the second bed frame. The housing corresponding to the first bed frame and the housing corresponding to the second bed frame are respectively connected to the ends of the first bed frame and the second bed frame.

[0296] Specifically, by simultaneously pressing the first handle and the second handle, a force is applied to the housings of the first handle and the second handle in the direction of movement of the positioning body corresponding to the preset positioning position. The housings of the first handle and the second handle are pressed against one side surface of the first bed frame and the second bed frame, respectively. This releases the limiting head of the limiting mechanism connected to the positioning body by the limiting mechanism configured on the first bed frame and the second bed frame, disengaging it from the limiting groove corresponding to the first bed frame and the second bed frame. While pressing the first handle and the second handle, the inner sides of the two palms press against one end of the two telescopic shafts, causing the two telescopic shafts to retract. The compression springs connected to the two telescopic shafts are then compressed, forcing the positioning fitting connected to the compression spring and capable of popping out of the bushing to pop out of the bushing, so that the positioning fitting can connect to the first positioning body corresponding to the preset positioning position (a "click" sound is heard at this time).

[0297] If the first positioning body corresponding to the preset positioning position is the specified preset positioning position, then the force applied to the housing of the first handle and the housing of the second handle along the direction of movement of the positioning body corresponding to the preset positioning position, as well as the force between the inner sides of the two palms and one end of the telescopic shaft, is released. At this time, the two telescopic shafts return from retracted to their natural state, and the compression springs connected to the telescopic shafts also return from retracted to their natural state. The positioning fitting part connected to the compression spring and capable of popping out the bushing retracts to the inner side of the bushing. The limiting head of the limiting mechanism connecting the limiting mechanism and the positioning body is inserted into the limiting grooves corresponding to the first bed frame and the second bed frame, realizing rapid assembly and positioning.

[0298] If the first positioning body corresponding to the preset positioning position is not the specified preset positioning position, then only the force between the inside of the palm and one end of the telescopic shaft is released. The positioning fitting part connected to the compression spring and capable of popping out the bushing retracts to the inside of the bushing. Under the force applied to the housing of the first handle along the direction of movement of the positioning body corresponding to the preset positioning position, it continues to move in the direction of movement of the positioning body corresponding to the next preset positioning position. After the positioning fitting part disengages from the first positioning body, the force between the inside of the palm and one end of the telescopic shaft continues until the preset positioning position is reached, thus achieving rapid assembly and positioning. If rapid release of positioning is achieved, only the force is applied to the housing of the first handle and the housing of the second handle (no force is applied to one end of the telescopic shaft). The limiting heads of the two limiting mechanisms and the limiting mechanisms connecting the positioning body disengage from the limiting slots corresponding to the first bed frame and the second bed frame, respectively.

[0299] Background and Purpose of this Disclosure: The Oxford Pathological Classification (MEST-C) of IgA nephropathy (IgAN) is the gold standard for assessing disease severity and prognosis. Tubular atrophy and interstitial fibrosis (MEST-T) reflects chronic irreversible damage, while endothelial cell proliferation (MEST-E) reflects acute active inflammation. Given the invasiveness and non-reproducibility of renal biopsy, this disclosure aims to explore the predictive efficacy of non-invasive biomarkers and clinical indicators for MEST-T and MEST-E pathological damage, construct corresponding diagnostic models, and evaluate their clinical application value in predicting longitudinal changes in eGFR and the benefit of immunosuppressive therapy. Methods of this Disclosure: Two independent retrospective cohorts of IgAN patients were included: a cross-sectional pathological prediction cohort (n=132) and a longitudinal treatment follow-up cohort (n=83). In the cross-sectional cohort, univariate logistic regression and receiver operating characteristic (ROC) curve analysis were used to assess the non-invasive diagnostic value of peripheral blood parameters (including creatinine, blood urea nitrogen, cystatin C, protein / albumin ratio [PAR], etc.) for MEST-T and MEST-E injuries, and internal cross-validation was performed using training and validation sets. In the follow-up cohort, with the dynamic changes in eGFR during the 1–2 year follow-up period as the primary outcome, multivariate regression analysis and interaction models were used to explore the differences in the response to immunosuppressive therapy among patients under different MEST-E risk stratifications (different injury grades of MEST-E).

[0300] Figure 2 The diagram illustrates the visualization of research results corresponding to embodiments of this disclosure; wherein, (a) the AUC values ​​of the non-invasive prognostic damage prediction model on the training and validation sets; (b) the ROC curve corresponding to the predicted MEST-T; (c) the ROC curve corresponding to the predicted MEST-E; (e) the PAR trajectory by PAR group and for those not receiving / receiving immunosuppressive therapy; and (f) the association between PAR changes and eGFR. Figure 2As shown, 1. The MEST-T (chronic pathological injury) non-invasive prediction model: Univariate analysis showed that baseline eGFR creatinine and blood urea nitrogen were the most significant independent correlation factors predicting MEST-T positivity. The MEST-T prediction model constructed based on this model exhibited extremely high and stable diagnostic performance, with an AUC of 0.902 on the training set and 0.900 / 0.904 on the validation set, indicating high sensitivity and specificity. 2. MEST-E (acute pathological inflammation) and PAR index: Peripheral blood PAR (protein / albumin ratio) was significantly correlated with MEST-E (independent samples t-test p = 0.0011). Compared to the normal group, peripheral blood PAR levels were significantly elevated in MEST-E positive patients (7.47 2.61 vs. 6.10 1.88). The AUC of the E model diagnostic diagnostic model based on PAR integrated indicators reached 0.777 in the training set and 0.770 / 0.774 in the validation set, suggesting that PAR is a sensitive non-invasive biomarker reflecting acute renal endothelial proliferation and active inflammation. 3. Longitudinal prognosis and immunosuppressive therapy response: Follow-up cohort analysis showed that in the low-inflammation and high-active-inflammation subgroups, patients receiving immunosuppressive therapy had a significantly greater increase in eGFR during the follow-up period than the untreated group. The above results indicate that the predictive model can highly accurately identify the degree of MEST-T chronic fibrosis in IgAN patients, and peripheral blood PAR has a good indicative role in MEST-E acute endothelial proliferation. The combined use of these two types of non-invasive models can not only assess the chronic and acute states of renal pathological damage, but also effectively guide the precise use of immunosuppressants and help clinicians identify IgAN patients in the reversible treatment window.

[0301] Those skilled in the art will understand that in the above-described method for non-invasive prediction of IgA nephropathy prognosis and determination of immunosuppressive therapy response in specific embodiments, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0302] According to one aspect of this disclosure, a non-invasive prediction system for prognostic damage in IgA nephropathy is provided, comprising: an acquisition unit for acquiring preset peripheral blood indicators corresponding to a cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, and preset peripheral blood indicators to be predicted; a construction unit for constructing a corresponding non-invasive prediction model for prognostic damage based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels; and a prediction unit for predicting prognostic damage based on the non-invasive prediction model for prognostic damage and the preset peripheral blood indicators to be predicted; or, comprising: a processor for storage and processing. The processor may include: a memory containing executable instructions; wherein the processor is configured to invoke instructions stored in the memory to execute the aforementioned non-invasive prediction method for prognostic damage in IgA nephropathy; or, a computer-readable storage medium storing a computer program / instructions and a bitstream thereon, wherein the computer program / instructions, when executed by a processor, implement the aforementioned non-invasive prediction method for prognostic damage in IgA nephropathy to generate the bitstream; or, a computer program product configured with a computer program / instructions, wherein the computer program / instructions, when executed by a processor, implement the aforementioned non-invasive prediction method for prognostic damage in IgA nephropathy.

[0303] According to one aspect of this disclosure, an immunosuppressive therapy response determination system is provided, comprising: an acquisition unit for acquiring preset peripheral blood indicators corresponding to a cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, and preset peripheral blood indicators to be predicted; a construction unit for constructing a corresponding non-invasive prognostic damage prediction model based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage labels; a first determination unit for determining a longitudinal treatment follow-up cohort corresponding to prognostic damage in the cross-sectional pathological prediction cohort based on the non-invasive prognostic damage prediction model and the preset peripheral blood indicators to be predicted; and a second determination unit for determining different outcomes based on an estimated glomerular filtration rate. Differences in response to immunosuppressive therapy under risk stratification of endothelial cell proliferation; or, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke instructions stored in the memory to perform the above-described immunosuppressive therapy response determination method; or, comprising: a computer-readable storage medium having stored thereon a computer program / instructions and a bit stream thereon, wherein the computer program / instructions, when executed by a processor, implement the above-described immunosuppressive therapy response determination method to generate the bit stream; or, comprising: a computer program product having configured computer programs / instructions, wherein the computer program / instructions, when executed by a processor, implement the above-described immunosuppressive therapy response determination method.

[0304] The various embodiments of this disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments.

Claims

1. A non-invasive method for predicting prognostic damage in IgA nephropathy, characterized in that, include: Obtain the preset peripheral blood indicators corresponding to the cross-sectional pathological prediction cohort and their corresponding IgA nephropathy prognostic pathological damage labels, as well as the preset peripheral blood indicators to be predicted. Based on the preset peripheral blood indicators and the prognostic pathological damage labels of IgA nephropathy, a corresponding non-invasive predictive model for prognostic damage is constructed. Based on the non-invasive prognostic damage prediction model and the preset peripheral blood indicators to be predicted, prognostic damage is predicted.

2. The non-invasive prediction method for prognostic damage in IgA nephropathy according to claim 1, characterized in that, Determining the preset peripheral blood indicators includes: acquiring multiple peripheral blood indicators to be processed corresponding to the cross-sectional pathological prediction cohort; performing logistic regression analysis on the multiple peripheral blood indicators to be processed and their corresponding IgA nephropathy prognostic pathological damage labels to obtain multiple levels of significance; configuring the peripheral blood indicators to be processed corresponding to the multiple levels of significance less than the preset level of significance as preset peripheral blood indicators; and / or, configuring the prognostic damage and the IgA nephropathy prognostic pathological damage labels to reflect renal tubular atrophy and interstitial fibrosis corresponding to chronic irreversible damage or glomerular endothelial cell hyperplasia corresponding to acute active inflammation.

3. The non-invasive prediction method for prognostic damage in IgA nephropathy according to any one of claims 1 or 2, characterized in that, The IgA nephropathy is configured as IgA nephropathy induced by tumor lesions; wherein, the prediction of IgA nephropathy induced by tumor lesions includes: using the quantitative imaging features of preset tumor lesions corresponding to the tumor lesions in medical images of preset locations at a first time point and the pathological results corresponding to induced or non-induced IgA nephropathy at a second time point after the first time point to train a preset predictor and construct a prediction model for IgA nephropathy induced by tumor lesions; using the prediction model for IgA nephropathy induced by tumor lesions and the quantitative imaging features of preset tumor lesions corresponding to the tumor lesions in medical images of preset locations to be predicted, to predict IgA nephropathy induced by tumor lesions.

4. The non-invasive prediction method for prognostic damage in IgA nephropathy according to claim 3, characterized in that, Before training the preset predictor, the method includes: extracting quantitative image features of tumor lesions corresponding to selected tumor lesions in medical images of preset locations; selecting preset quantitative image features of tumor lesions related to the pathological results from the selected quantitative image features of tumor lesions; and / or determining preset quantitative image features of tumor lesions related to the pathological results, including: using a univariate logistic regression analysis algorithm, the selected quantitative image features of tumor lesions and their corresponding pathological results, performing univariate logistic regression on each of the selected quantitative image features of tumor lesions to determine the preset quantitative image features of tumor lesions related to the pathological results.

5. The non-invasive prediction method for prognostic damage in IgA nephropathy according to any one of claims 3 or 4, characterized in that, The quantitative imaging features of the tumor lesions to be selected include one or more of the following: the long diameter of the tumor lesion, the long diameter of the solid component of the tumor lesion, the ratio of the solid component of the tumor lesion to the total tumor lesion, the location of the tumor lesion, whether the edge of the tumor lesion is clear, the morphology of the tumor lesion, the spinous process sign of the tumor lesion, the lobulation sign of the tumor lesion, the pleural traction sign of the tumor lesion, the cavitation sign of the tumor lesion, and the air bronchus sign of the tumor lesion.

6. The non-invasive prediction method for prognostic damage in IgA nephropathy according to claim 5, characterized in that, Determining whether the edge of the tumor lesion corresponding to the quantitative imaging features of the selected tumor lesion is clear includes: performing an erosion operation on the tumor lesion mask image to obtain a tumor lesion mask eroded image; subtracting the tumor lesion mask eroded image from the tumor lesion mask image to obtain a tumor lesion mask processed image; mapping the tumor lesion location coordinates in the tumor lesion mask processed image to the preset site medical image; calculating multiple gradients corresponding to the tumor lesion location coordinates in the preset site medical image; if the gradient magnitude values ​​corresponding to the multiple tumor lesions are greater than or equal to a preset gradient magnitude value, then the boundary of the tumor lesion location corresponding to the tumor lesion location greater than or equal to the preset gradient magnitude value is determined to be clear; otherwise, the tumor lesion boundary is determined to be unclear.

7. The non-invasive prediction method for prognostic damage in IgA nephropathy according to any one of claims 5 or 6, characterized in that, Determining the spur includes: detecting the edges of the tumor lesion mask in the tumor lesion mask image to obtain the tumor lesion mask edge line; calculating the first pixel distance and the second pixel distance between any tumor lesion mask edge pixel on the tumor lesion mask edge line and its adjacent tumor lesion mask edge pixels; if the first pixel distance and the second pixel distance are greater than a set pixel distance, then configuring the arbitrary tumor lesion mask edge pixel as a spur; and / or determining the lobulation sign of the tumor lesion includes: performing an erosion operation on the tumor lesion mask image corresponding to each tumor lesion in the medical image of the preset location to obtain a tumor lesion mask erosion image; subtracting the corresponding tumor lesion mask erosion image from the tumor lesion mask image of each tumor lesion to obtain a tumor lesion mask processing image corresponding to each tumor lesion; judging the morphology of the tumor lesion in the tumor lesion mask processing image corresponding to each tumor lesion; if the morphology of the tumor lesion is irregular, then determining whether there is a tumor lesion lobulation sign in the tumor lesion mask processing image corresponding to each tumor lesion.

8. A method for determining the response to immunosuppressive therapy, characterized in that, include: Using the non-invasive prediction method for prognostic damage in IgA nephropathy as described in any one of claims 1-7, determine the longitudinal treatment follow-up cohort corresponding to the prognostic damage in the cross-sectional pathological prediction cohort; The estimated glomerular filtration rate was used as the outcome to determine the differences in response to immunosuppressive therapy under different risk stratifications of endothelial cell proliferation.

9. A non-invasive predictive system for prognostic damage in IgA nephropathy, characterized in that, include: The acquisition unit is used to acquire the preset peripheral blood indicators corresponding to the cross-sectional pathological prediction queue and their corresponding IgA nephropathy prognostic pathological damage labels, as well as the preset peripheral blood indicators to be predicted. The construction unit is used to construct a corresponding non-invasive prognostic damage prediction model based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage label; the prediction unit is used to predict prognostic damage based on the non-invasive prognostic damage prediction model and the preset peripheral blood indicators to be predicted. or, The method comprises: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the non-invasive prediction method for prognostic damage in IgA nephropathy according to any one of claims 1-8; or, a computer-readable storage medium having a computer program / instructions and a bit stream stored thereon, wherein the computer program / instructions, when executed by a processor, implement the non-invasive prediction method for prognostic damage in IgA nephropathy according to any one of claims 1-8 to generate the bit stream; or, a computer program product having a computer program / instructions configured to implement the non-invasive prediction method for prognostic damage in IgA nephropathy according to any one of claims 1-8 when executed by a processor.

10. A system for determining the response to immunosuppressive therapy, characterized in that, include: The acquisition unit is used to acquire the preset peripheral blood indicators corresponding to the cross-sectional pathological prediction queue and their corresponding IgA nephropathy prognostic pathological damage labels, as well as the preset peripheral blood indicators to be predicted. The construction unit is used to construct a corresponding non-invasive prediction model for prognostic damage based on the preset peripheral blood indicators and the IgA nephropathy prognostic pathological damage label; the first determination unit is used to determine the longitudinal treatment follow-up cohort corresponding to the prognostic damage in the cross-sectional pathological prediction cohort based on the non-invasive prediction model for prognostic damage and the preset peripheral blood indicators to be predicted. The second determining unit is used to determine the differences in response to immunosuppressive therapy under different risk stratifications of endothelial cell proliferation, with the estimated glomerular filtration rate as the outcome. or, The method comprises: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke instructions stored in the memory to execute the immunosuppressive therapy response determination method of claim 9; or, a computer-readable storage medium having a computer program / instructions and a bit stream stored thereon, wherein the computer program / instructions, when executed by a processor, implement the immunosuppressive therapy response determination method of claim 9 to generate the bit stream; or, a computer program product having a computer program / instructions configured thereon, wherein the computer program / instructions, when executed by a processor, implement the immunosuppressive therapy response determination method of claim 9.

Images