Classifier-based medical ultrasonic data classification method and device

Abstract

The invention relates to the technical field of medical care informatics, in particular to a medical ultrasonic data classification method and device based on a classifier, and the method comprises the steps: obtaining a plurality of times of medical ultrasonic data including the current medical ultrasonic data of a tumor patient; dividing the treatment process of the tumor patient into a plurality of treatment stages based on the detection time of the medical ultrasonic data; based on the distribution stability condition of the detection time interval of the adjacent medical ultrasonic data in the treatment stage and the distribution condition of different tumor related indexes in the medical ultrasonic data in the treatment stage, determining a tumor dynamic influence comprehensive evaluation value in the treatment stage; and based on the tumor dynamic influence comprehensive evaluation value of the treatment stage to which the current medical ultrasonic data belongs, classifying the current medical ultrasonic data by using a classifier. According to the method, the tumor dynamic influence comprehensive evaluation value of each treatment stage is determined, so that the classification accuracy of the medical ultrasonic data is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care informatics, and in particular to a method and device for classifying medical ultrasound data based on a classifier. Background Art

[0002] With the continuous advancement of medical technology, medical ultrasound data classification has become a crucial component of data processing in the healthcare industry. Accurately classifying medical ultrasound data through automated and intelligent methods, particularly for tumors and cardiovascular disease, helps doctors quickly detect early symptoms of disease and develop intervention strategies based on the classification results, thereby improving patient survival rates and quality of life.

[0003] In existing technologies, medical ultrasound data is typically classified using classifiers. Classifiers are core components that automatically identify and classify lesions (such as tumors and nodules) in ultrasound images. Their core function is to map extracted ultrasound image features to predefined disease categories (e.g., benign / malignant, different subtypes, etc.). Existing classifiers include support vector machines (SVMs) and convolutional neural networks (CNNs). Because existing classifiers assess and classify based on the quality of ultrasound images, they ignore the dynamic temporal evolution of the disease, which can affect image quality, when classifying medical ultrasound data from cancer patients. For example, rapid tumor infiltration and growth can lead to increased boundary blurring. Furthermore, classifiers cannot effectively capture the dynamic changes in tumor manifestations at different stages, resulting in inaccurate classification of medical ultrasound data. Summary of the Invention

[0004] In order to solve the above-mentioned technical problem of inaccurate classification of medical ultrasound data, the purpose of the present invention is to provide a method and device for classifying medical ultrasound data based on a classifier. The technical solution adopted is as follows: In a first aspect, the present invention provides a method for classifying medical ultrasound data based on a classifier, comprising the following steps: Acquiring several medical ultrasound data of a tumor patient including current medical ultrasound data; dividing the treatment process of the tumor patient into several treatment stages based on the detection time of all the medical ultrasound data; Determining a comprehensive evaluation value of the dynamic impact of the tumor during the treatment phase based on a distribution stability of detection time intervals between adjacent medical ultrasound data during the treatment phase and a distribution of different tumor-related indicators in all the medical ultrasound data during the treatment phase; The current medical ultrasound data is classified using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs.

[0005] In conjunction with the first aspect above, in some possible implementations, the treatment process of the tumor patient is divided into several treatment stages, including: determining a time interval sequence of the tumor patient based on time intervals between detection times of adjacent medical ultrasound data; dividing the detection time interval sequence into a plurality of subsequences based on a change trend of a local detection time interval in the detection time interval sequence; A number of treatment stages of the treatment process of the tumor patient are determined based on the number of ultrasound detections corresponding to the detection time intervals in the subsequence.

[0006] In conjunction with the first aspect above, in some possible implementations, the different tumor-related indicators include at least tumor volume, blood flow resistance index, and number of calcification points; and determining the comprehensive assessment value of the dynamic impact of the tumor during the treatment stage includes: determining a stability evaluation coefficient of the treatment stage based on a distribution stability of detection time intervals between adjacent medical ultrasound data in the treatment stage; determining the controllability of tumor manifestation in the treatment phase based on changes in tumor volume and blood flow resistance index in all the medical ultrasound data in the treatment phase; determining the tumor malignant invasion impact degree during the treatment phase based on changes in the number of calcification points and blood flow resistance index in all the medical ultrasound data during the treatment phase; Based on the stability evaluation coefficient, the controllability of the tumor performance and the influence of the tumor malignant invasion are weightedly fused to determine a comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage.

[0007] In conjunction with the first aspect above, in some possible implementations, determining the stability assessment coefficient of the treatment stage includes: determining a standard deviation of detection time intervals of all adjacent medical ultrasound data in the treatment phase to obtain a time interval dispersion; A negative correlation normalization process is performed on the time interval dispersion to obtain a stable evaluation coefficient of the treatment stage.

[0008] In conjunction with the first aspect above, in some possible implementations, determining the degree of controllability of tumor manifestations during the treatment phase includes: determining a controllable degree of tumor volume change based on a change rate of the tumor volume in all the medical ultrasound data during the treatment phase; determining the true growth representation of the tumor based on the dispersion of the blood flow resistance index in all the medical ultrasound data during the treatment phase; The actual tumor growth expression degree is used as a weight to perform weighted multiplication and normalization on the tumor volume change controllability to obtain the tumor expression controllability at the treatment stage.

[0009] In conjunction with the first aspect above, in some possible implementations, determining the tumor malignant invasion impact at the treatment stage includes: determining a calcification growth rate expression degree based on the number of calcification points in all the medical ultrasound data and the detection time of all the medical ultrasound data during the treatment stage; determining a steep rise degree of the blood resistance index based on changes in the blood resistance index in all the medical ultrasound data during the treatment phase; The tumor malignant invasion impact degree in the treatment stage is calculated based on the calcification growth rate expression degree and the blood resistance index steepness.

[0010] In conjunction with the first aspect above, in some possible implementations, determining the calcification growth rate expression level includes: determining a maximum number of calcification points among all the calcification points in the medical ultrasound data during the treatment phase; determining a time interval between the medical ultrasound data corresponding to the maximum number of calcification points and a detection time of the first medical ultrasound data in the treatment stage as a target time interval; Determining the ratio of the target time interval to the entire time interval spanned by the treatment phase as the maximum calcification penetration; Normalizing the ratio of the maximum number of calcification points to the number of calcification points in the medical ultrasound data detected by the first ultrasound examination during the treatment phase to obtain a maximum calcification expression; The maximum calcification expression penetration is multiplied by the maximum calcification expression to obtain the calcification growth rate expression.

[0011] In conjunction with the first aspect above, in some possible implementations, determining the steepness of the blood resistance index includes: determining target medical ultrasound data corresponding to a maximum number of calcification points among all the medical ultrasound data in the treatment stage; determining a maximum ratio of a blood resistance index in a subsequent medical ultrasound data of the target medical ultrasound data to a blood resistance index in each preceding medical ultrasound data; The maximum ratio is normalized to obtain the steep rise degree of the blood resistance index.

[0012] In combination with the first aspect above, in some possible implementations, the classifier is a decision tree; and classifying the current medical ultrasound data includes: Obtaining various quality indicators in the current medical ultrasound data; Determining a maximum value among the comprehensive evaluation values ​​of dynamic tumor impacts of the treatment stage to which the current medical ultrasound data belongs, and obtaining a final comprehensive evaluation value of dynamic tumor impacts of the current medical ultrasound data; Various quality indicators of the current medical ultrasound data and the final comprehensive evaluation value of the dynamic impact of the tumor are input into a pre-built decision tree, and the current medical ultrasound data is classified using the decision tree to obtain a classification result.

[0013] In a second aspect, the present invention further provides a classifier-based medical ultrasound data classification device, comprising a memory and a processor. The memory is configured to store executable computer program code, and the processor is configured to retrieve and execute the executable computer program code from the memory, so that the system performs the method of the first aspect or any possible implementation of the first aspect.

[0014] In a third aspect, the present invention further provides a classification system for medical ultrasound data based on a classifier, the system comprising: A data acquisition module, used to acquire several medical ultrasound data of a tumor patient including the current medical ultrasound data; a stage division module, configured to divide the treatment process of the tumor patient into a plurality of treatment stages based on the detection time of all the medical ultrasound data; an evaluation module, configured to determine a comprehensive evaluation value of the dynamic impact of the tumor during the treatment phase based on a distribution stability of detection time intervals between adjacent medical ultrasound data during the treatment phase and a distribution of different tumor-related indicators in all the medical ultrasound data during the treatment phase; The classification module is used to classify the current medical ultrasound data using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs.

[0015] In a fourth aspect, the present invention also provides a computer program product, which includes: computer program code, which, when running on a computer, enables the computer to execute a classification method for medical ultrasound data based on a classifier in the above-mentioned first aspect or any possible implementation of the first aspect.

[0016] In a fifth aspect, the present invention also provides a computer-readable storage medium, which stores a computer program code. When the computer program code runs on a computer, the computer executes a classification method for medical ultrasound data based on a classifier in the above-mentioned first aspect or any possible implementation of the first aspect.

[0017] The present invention has the following beneficial effects: by acquiring several medical ultrasound data of a tumor patient, including the current medical ultrasound data, and dividing the tumor patient's treatment process into multiple treatment stages based on the detection time of the medical ultrasound data; analyzing the distribution stability of the detection time intervals of adjacent medical ultrasound data in each treatment stage, as well as the distribution of different tumor-related indicators in the medical ultrasound data, analyzing the dynamic development of the tumor in each stage, and determining a comprehensive evaluation value of the dynamic impact of the tumor in each treatment stage; and thereby classifying the current medical ultrasound data using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs. By analyzing the dynamic development changes of the tumor of a tumor patient in each treatment stage, the present invention improves the classifier's ability to identify tumor ultrasound data of different stages, thereby ultimately improving the classification accuracy of the medical ultrasound data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is a flowchart of a method for classifying medical ultrasound data based on a classifier according to an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a decision tree according to an embodiment of the present invention; Figure 3 2 is a schematic structural diagram of a medical ultrasound data classification system based on a classifier according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to clearly illustrate the technical features of this solution, the present invention is described in detail below through specific implementation methods in conjunction with the accompanying drawings.

[0021] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0022] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0023] As used herein, the term "including" and its variations are open-ended, i.e., "including but not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment," the term "another embodiment" means "at least one additional embodiment," and the term "some embodiments" means "at least some embodiments." Other terms are defined in the following description.

[0024] It should be noted that the concepts of "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0025] Although operations or steps are described in a particular order in the drawings in the embodiments of the present invention, this should not be understood as requiring that these operations or steps be performed in the particular order shown or in a serial order, or that all of the operations or steps shown be performed to obtain a desired result. In the embodiments of the present invention, these operations or steps may be performed serially; they may also be performed in parallel; or a portion of these operations or steps may be performed.

[0026] At the same time, it is understood that the data involved in the technical solutions of the present invention (including but not limited to the data itself, the acquisition or use of the data) must comply with the requirements of relevant laws, regulations and relevant provisions. Unless otherwise defined, all technical and scientific terms used in this invention have the same meanings as those commonly understood by those skilled in the art to which this invention belongs, and all parameters or indicators in the formulas involved in this invention are normalized values ​​to eliminate dimension effects.

[0027] In order to solve the problem of inaccurate classification of existing medical ultrasound data, an embodiment of the present invention provides a classification method and device for medical ultrasound data based on a classifier. By analyzing several medical ultrasound data of tumor patients, the treatment process of tumor patients is divided into several treatment stages, and the comprehensive evaluation value of the dynamic impact of the tumor in each treatment stage is determined. Based on the comprehensive evaluation value of the dynamic impact of the tumor, the current medical ultrasound data of the tumor patient is classified, effectively improving the classification accuracy of the medical ultrasound data.

[0028] A method and apparatus for classifying medical ultrasound data based on a classifier provided by an embodiment of the present invention will be described in detail below with reference to the accompanying drawings.

[0029] Figure 1 FIG. 1 shows a basic flow chart of a method for classifying medical ultrasound data based on a classifier provided by an embodiment of the present invention. Figure 1 As shown, the method specifically includes the following steps: Step S100: Acquire several medical ultrasound data of a tumor patient including the current medical ultrasound data; Step S200: Dividing the treatment process of the tumor patient into several treatment stages based on the detection time of all the medical ultrasound data; Step S300: determining a comprehensive evaluation value of the dynamic impact of the tumor during the treatment phase based on the distribution stability of the detection time intervals between adjacent medical ultrasound data during the treatment phase and the distribution of different tumor-related indicators in all the medical ultrasound data during the treatment phase; Step S400: Classifying the current medical ultrasound data using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs.

[0030] In the classifier-based medical ultrasound data classification method provided in the above-mentioned embodiment of the present invention, the treatment process of tumor patients is divided into several treatment stages corresponding to different stages of disease development, and the dynamic development characteristics of the tumor over time in each treatment stage are quantified, and the comprehensive evaluation value of the dynamic impact of the tumor in each treatment stage is determined. Based on the comprehensive evaluation value of the dynamic impact of the tumor, the current medical ultrasound data of the tumor patient is classified, breaking through the limitations of static image quality evaluation and effectively improving the classification accuracy of medical ultrasound data.

[0031] The following is a detailed introduction to each step of a classifier-based medical ultrasound data classification method provided by an embodiment of the present invention.

[0032] Step S100: Acquire several medical ultrasound data of a tumor patient including current medical ultrasound data.

[0033] Specifically, for diseases that develop and evolve dynamically, such as tumors (such as thyroid cancer, breast cancer, liver cancer, etc.), tumor patients usually require long-term monitoring and treatment. During the long-term treatment process, tumor patients need to undergo multiple ultrasound examinations. Therefore, when it is necessary to classify the current medical ultrasound data of tumor patients, in order to avoid the errors caused by classification based only on the quality of the film of a single test result, multiple medical ultrasound data of the tumor patient's historical treatment process can be obtained. Based on multiple medical ultrasound data, the current medical ultrasound data can be refined and classified from the long-term analysis level such as the tumor patient's deterioration trend or recovery progress, thereby improving the accuracy of ultrasound data classification.

[0034] In a specific example, based on the electronic medical record information of a tumor patient, the current medical ultrasound data of the tumor patient and multiple medical ultrasound data of historical ultrasound detections are obtained, thereby obtaining several medical ultrasound data of the tumor patient including the current medical ultrasound data, and the medical ultrasound data is specifically image data.

[0035] Step S200: Dividing the treatment process of the tumor patient into several treatment stages based on the detection time of all the medical ultrasound data.

[0036] Specifically, since cancer patients usually require long-term monitoring and treatment, different cancer patients will be in relatively different treatment stages. For cancer patients themselves, they usually undergo corresponding chemotherapy, drug treatment and other processes after the tumor is diagnosed, and use ultrasound examinations to reflect specific related tumor manifestations. Therefore, the number of ultrasound examinations performed on cancer patients can be used as reference data to reflect their relative treatment progress, thereby dividing the treatment process into different treatment stages.

[0037] Furthermore, in a possible implementation, dividing the treatment process of the tumor patient into several treatment stages includes: determining the time interval sequence of the tumor patient based on the time interval between the detection times of adjacent medical ultrasound data; dividing the detection time interval sequence into several subsequences based on the change trend of the local detection time interval in the detection time interval sequence; and determining the several treatment stages of the treatment process of the tumor patient based on the number of ultrasound detections corresponding to the detection time intervals in the subsequences.

[0038] In a specific example, first, based on the electronic medical record information of the tumor patient, the time of each ultrasound examination of the tumor patient is counted, so as to obtain the time interval (in days) between adjacent ultrasound examinations.

[0039] Secondly, based on the time intervals between adjacent ultrasound examinations of the tumor patient, all time intervals are arranged in time sequence to obtain a time interval sequence of the tumor patient.

[0040] Finally, since the more stable the time intervals between adjacent ultrasound examinations, the more likely the patient is in the maintenance treatment phase, while shorter intervals indicate the likelihood of tumor recurrence or an intensive efficacy evaluation phase, we determined a subsequence of consecutive time intervals within the interval sequence that showed a continuously decreasing pattern, and the corresponding number of ultrasound examinations was used as the intensive efficacy evaluation phase for the patient. We also determined a subsequence of consecutive time intervals that showed both increasing and constant patterns within the interval sequence, and the corresponding number of ultrasound examinations was used as the maintenance treatment phase for the patient. For example, a patient underwent eight ultrasound examinations, corresponding to the time intervals [60, 61, 62, 30, 14, 15, 14]. The time intervals corresponding to the maintenance treatment phase were [60, 61, 62] and [14, 15], respectively, and the corresponding number of ultrasound examinations were [1, 2, 3, 4] and [5, 6, 7], respectively. The time intervals corresponding to the intensive efficacy evaluation phase were [62, 30, 14] and [15, 14], respectively, and the corresponding number of ultrasound examinations were [3, 4, 5, 6] and [6, 7, 8], respectively.

[0041] In this way, the above step S200 can determine several treatment stages of the tumor patient based on the detection time of multiple medical ultrasound data of the tumor patient, each treatment stage being an intensive efficacy evaluation stage or a maintenance treatment stage.

[0042] Step S300: Determine a comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage based on the distribution stability of the detection time intervals of adjacent medical ultrasound data in the treatment stage and the distribution of different tumor-related indicators in all the medical ultrasound data in the treatment stage.

[0043] Specifically, based on the electronic medical record information of cancer patients, different tumor-related indicators are obtained from the medical ultrasound data of each ultrasound examination of the cancer patients. The different tumor-related indicators can specifically include parameters such as tumor volume (through the three-diameter method), blood flow resistance index (through the Doppler spectrum envelope calculation method), and calcification point count (through high echo point counting).

[0044] For cancer patients categorized into various treatment stages, the impact of fluctuations in time intervals within each stage must be fully considered. More stable time intervals indicate that tumor growth or changes are more controllable, and the patient's condition remains relatively stable during this period. Conversely, less stable time intervals indicate more volatile tumor changes and a greater likelihood of sudden changes. Therefore, based on the stability of the distribution of time intervals between adjacent medical ultrasound data during treatment, a stability assessment coefficient for cancer patients at each treatment stage can be determined.

[0045] At the same time, for cancer patients, their tumors may expand or become relatively stable due to relevant medical interventions during different treatment stages. For example, in some early stages of treatment, when tumor cells develop resistance to treatment (classified as primary resistance or natural resistance) or the treatment method fails to effectively inhibit tumor growth, the tumor may continue to expand through mechanisms such as proliferation and angiogenesis. If the treatment is effective, the tumor may enter a relatively stable state, meaning that the size and growth rate of the tumor remain unchanged or change slightly. Therefore, based on the distribution of tumor-related indicators (such as tumor volume) in medical ultrasound data during the treatment stage, the tumor development status of cancer patients can be evaluated at each treatment stage to determine the controllability of tumor performance at each treatment stage, reflecting the degree to which the changes in tumor size and growth are controlled by the relevant treatment during that treatment stage.

[0046] Furthermore, due to the development of the patient's own tumor and the influence of therapeutic interventions, even a tumor in a relatively controllable treatment state may still develop malignant effects due to disease progression. In other words, even if the tumor does not significantly increase in size, it may still cause the condition to worsen. Tumor progression often manifests as a strong malignant invasiveness, with cancer cells having a strong ability to break through the basement membrane and invade surrounding tissues. This is often accompanied by changes in other related indicators. For example, during tumor progression, changes such as angiogenesis and tumor necrosis often occur, thereby increasing the number of calcifications. For cancer patients receiving treatment, a reduction in tumor volume is often accompanied by a decrease in the number of calcifications, indicating that treatment has achieved positive results in inhibiting tumor growth and improving the tumor microenvironment. Therefore, the degree of tumor malignant invasiveness at each treatment stage can be determined based on the distribution of tumor-related indicators (such as the number of calcifications) in medical ultrasound data during the treatment phase.

[0047] Furthermore, by comprehensively considering the stability evaluation coefficient, tumor manifestation controllability and tumor malignant invasion influence of cancer patients at each treatment stage, the comprehensive evaluation value of tumor dynamic impact at each treatment stage is determined to reflect the tumor development trend of cancer patients at different treatment stages.

[0048] Furthermore, in a possible implementation, determining the comprehensive evaluation value of the dynamic impact of the tumor during the treatment phase includes: Step S301: determining a stability evaluation coefficient of the treatment stage based on a distribution stability of detection time intervals between adjacent medical ultrasound data in the treatment stage.

[0049] The stability of the time intervals between ultrasound examinations during each treatment phase was analyzed to determine a stability assessment coefficient for each treatment phase. A more stable distribution of time intervals indicates that the tumor's growth or change trends are more controllable and the patient's condition remains relatively stable during this period, resulting in a higher stability assessment coefficient for that treatment phase.

[0050] Furthermore, in a possible implementation, determining the stability evaluation coefficient of the treatment stage includes: determining the standard deviation of the detection time intervals of all adjacent medical ultrasound data in the treatment stage to obtain the time interval dispersion; performing negative correlation normalization processing on the time interval dispersion to obtain the stability evaluation coefficient of the treatment stage.

[0051] In a specific example, the standard deviation of all time intervals corresponding to each treatment stage is determined to obtain the time interval dispersion, and the time interval dispersion is inversely normalized, that is, the reciprocal of the sum of the time interval dispersion and a minimum value greater than 0 (to prevent the denominator from being 0) is taken, and the reciprocal, that is, the normalized value, is used as the stability evaluation coefficient of the treatment stage.

[0052] Step S302: determining the controllability of tumor manifestation in the treatment phase based on changes in tumor volume and blood flow resistance index in all the medical ultrasound data in the treatment phase.

[0053] Based on changes in tumor volume in medical ultrasound data during each treatment phase, the degree of control of tumor volume changes during that phase, influenced by physician diagnosis and medication intervention, can be analyzed. However, changes in tumor volume during a treatment phase do not necessarily reflect true tumor growth. This is because factors such as inflammation, edema, and vasodilation may cause the tumor to appear larger on imaging, but this does not represent actual tumor growth. Therefore, assessing tumor control solely based on tumor volume changes is overly simplistic. Instead, the blood flow resistance index is needed to assess the blood supply to the tumor and surrounding tissues. An increase in the blood flow resistance index may indicate increased tumor malignancy or growth, while a decrease in the blood flow resistance index may be related to tumor neovascularization, indicating true growth. A stable blood flow resistance index suggests that the increase in tumor volume is not due to tumor growth but rather a false increase caused by other factors. Therefore, by jointly analyzing changes in tumor volume and the blood flow resistance index in the medical ultrasound data during each treatment phase, the true degree of tumor control can be determined at each treatment phase.

[0054] Furthermore, in a possible implementation, determining the controllability of tumor manifestation in the treatment stage includes: determining the controllability of tumor volume change based on the rate of change of tumor volume in all the medical ultrasound data in the treatment stage; determining the true growth manifestation of the tumor based on the degree of dispersion of the blood flow resistance index in all the medical ultrasound data in the treatment stage; and using the true growth manifestation of the tumor as a weight to perform weighted multiplication and normalization on the controllability of tumor volume change to obtain the controllability of tumor manifestation in the treatment stage.

[0055] In a specific example, for each treatment stage, the tumor volume and the time (in months) between any adjacent ultrasound examinations (n+1) and nth are calculated, respectively, as the volume difference and time difference between the n+1 and nth ultrasound examinations. The volume difference and time difference between the n+1 and nth ultrasound examinations are then compared to determine the tumor volume change rate between the n+1 and nth ultrasound examinations in each treatment stage. The difference between the i-th tumor volume change rate and the i+1-th tumor volume change rate in each treatment stage is then calculated. All differences across stages are accumulated, and the accumulated value is recorded as the tumor volume change controllability for each treatment stage. A higher value for the tumor volume change controllability indicates a higher degree of tumor control and lower tumor activity.

[0056] Next, the standard deviation of the blood flow resistance index across all ultrasound scans within each treatment phase was determined and recorded as the true tumor growth representation for each treatment phase. The true tumor growth representation for each treatment phase was used as a weight to reflect the true tumor growth during that treatment phase, as reflected by changes in the blood flow resistance index, rather than changes in tumor volume due to other factors.

[0057] Finally, the product of the actual tumor growth expression and the controllability of tumor volume changes at each treatment stage was calculated, and the product was normalized using the norm normalization function. The normalized value was recorded as the tumor expression controllability at each treatment stage, that is, the actual controllability of tumor volume changes. The actual controllability of tumor volume changes reflects the degree to which tumor size growth changes can be controlled by relevant treatments at each treatment stage.

[0058] Step S303: determining the tumor malignant invasion impact degree in the treatment stage based on the changes in the number of calcification points and the blood flow resistance index in the medical ultrasound data in the treatment stage.

[0059] Changes in the number of calcifications are often associated with the malignancy of the tumor. A higher number of calcifications often indicates a more aggressive tumor. Furthermore, as the tumor grows in size, the number of calcifications generally increases. Therefore, changes in calcifications not only reflect the growth status of the tumor but also provide important clues for assessing its malignancy.

[0060] Calcification and blood flow also have a biological connection in tumor progression. Abnormal calcification often occurs around necrotic areas, while high blood flow reflects the pathological state of neovascularization. Together, they indicate a disturbed tumor microenvironment. Therefore, an increasing rate of calcification can serve as a trigger for malignant transformation, while a sharp rise in the blood resistance index can signal accelerated invasiveness. When both calcification and blood flow resistance become abnormal, it indicates that the tumor's malignant biological behavior has crossed a critical point.

[0061] Furthermore, in a possible implementation, determining the influence of tumor malignant invasion in the treatment stage includes: determining the calcification growth rate expression based on the number of calcification points in all the medical ultrasound data and the detection time of all the medical ultrasound data in the treatment stage; determining the steepness of the blood resistance index based on the change of the blood flow resistance index in all the medical ultrasound data in the treatment stage; and calculating the influence of tumor malignant invasion in the treatment stage based on the calcification growth rate expression and the steepness of the blood resistance index.

[0062] Among them, determining the calcification growth rate expression includes: determining the maximum number of calcification points among all the medical ultrasound data in the treatment stage; determining the time interval between the medical ultrasound data corresponding to the maximum number of calcification points and the detection time of the first medical ultrasound data in the treatment stage as the target time interval; determining the ratio of the target time interval to the overall time interval spanning the treatment stage as the maximum calcification expression penetration; normalizing the ratio of the maximum number of calcification points to the number of calcification points in the medical ultrasound data of the first ultrasound detection in the treatment stage to obtain the maximum calcification expression; multiplying the maximum calcification expression penetration by the maximum calcification expression to obtain the calcification growth rate expression.

[0063] Determining the steepness of the blood resistance index includes: determining the target medical ultrasound data corresponding to the maximum number of calcification points among all the medical ultrasound data in the treatment stage; determining the maximum ratio of the blood resistance index in the subsequent medical ultrasound data of the target medical ultrasound data to the blood resistance index in the previous medical ultrasound data; and normalizing the maximum ratio to obtain the steepness of the blood resistance index.

[0064] In one specific example, the number of calcification points in the medical ultrasound data from each ultrasound examination at each treatment stage, as well as the total time interval within the overall time interval of that examination stage, are first obtained. The time interval between the ultrasound examination with the maximum number of calcification points and the first ultrasound examination at that examination stage is marked as the target time interval. The target time interval between the ultrasound examination with the maximum number of calcification points and the first ultrasound examination at that stage is then compared with the total time interval within the overall time interval of that stage to determine the maximum calcification penetration at each treatment stage.

[0065] The maximum number of calcification points is compared with the number of calcification points in the medical ultrasound data of the initial ultrasound examination in the detection stage, and the ratio obtained by the comparison is normalized using the norm normalization function. The normalized value is used as the maximum calcification expression degree in each treatment stage.

[0066] The calcification growth rate is reflected by whether the maximum calcification expression at each treatment stage indicates a continuous increase in the number of calcification points throughout the entire treatment stage. The calcification growth rate for each treatment stage is calculated by multiplying the maximum calcification expression penetration by the maximum calcification expression.

[0067] Secondly, in the early stage of tumor malignancy, the increase in the number of microcalcifications precedes obvious blood flow abnormalities. Calcification is a product of abnormal cellular metabolism, and angiogenesis requires time. Therefore, the assessment of the steep rise in the blood resistance index can be based on the analysis of the blood resistance index of multiple ultrasound examinations between the last ultrasound examination corresponding to the maximum calcification point in the current stage and the initial ultrasound examination in this treatment stage.

[0068] Therefore, the blood resistance index in the medical ultrasound data of the last ultrasound detection corresponding to the maximum number of calcification points in each treatment stage is compared with the blood resistance index in the medical ultrasound data of the previous ultrasound detections, and the maximum ratio is selected and normalized using the norm normalization function. The normalized value is used as the steepness of the blood resistance index in each treatment stage.

[0069] Finally, the steep rise of the blood resistance index and the expression of calcification growth rate in each treatment stage were combined to determine the impact of tumor malignant invasion in each treatment stage. That is, the steep rise of the blood resistance index and the expression of calcification growth rate in each treatment stage were multiplied, and the product was determined as the impact of tumor malignant invasion in each treatment stage.

[0070] Step S304: Based on the stability evaluation coefficient, weighted fusion is performed on the tumor manifestation controllability and the tumor malignant invasion influence to determine a comprehensive evaluation value of the tumor dynamic influence in the treatment stage.

[0071] Based on the stability assessment coefficient of each treatment stage, the weighted values ​​of the controllability of tumor manifestations and the influence of tumor malignant invasion in that treatment stage are determined. When the stability assessment coefficient is larger, it means that the tumor growth is more likely to be under control, and at this time, we should focus on further controlling tumor growth; when the stability assessment coefficient is smaller, it means that the tumor is more malignant and invasive, and more radical intervention measures are needed to prevent tumor expansion and metastasis.

[0072] In a specific example, the stability evaluation coefficient for each treatment stage is recorded as ,Will As the weighted value of the tumor malignant invasion influence, As the weighted value of the controllability of tumor performance, the comprehensive evaluation value of tumor dynamic impact at each treatment stage is calculated using the following formula: ; in, It represents the comprehensive evaluation value of tumor dynamic impact at each treatment stage; Indicates the impact of tumor malignancy and invasion at each treatment stage, Indicates the degree of tumor control at each treatment stage.

[0073] The above-mentioned step S300 can accurately calculate the comprehensive evaluation value of the dynamic impact of the tumor in each treatment stage by combining the distribution stability of the detection time intervals of adjacent medical ultrasound data in each treatment stage, and combining the distribution of tumor volume, blood flow resistance index and calcification point number in the medical ultrasound data in each treatment stage. The comprehensive evaluation value of the dynamic impact of the tumor reflects the degree of tumor development trend in different stages.

[0074] Step S400: Classifying the current medical ultrasound data using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs.

[0075] Specifically, the current classification of medical ultrasound data involves extracting various quality indicators from the data and constructing a decision tree based on these indicators. To improve the accuracy of medical ultrasound data classification, the original decision tree was rebuilt by incorporating the comprehensive assessment of the dynamic impact of the tumor at the treatment stage of the current medical ultrasound data (the most recent medical ultrasound data) from multiple medical ultrasound data sets as a classification indicator at a certain level in the decision tree. This reconstructed decision tree then accurately classifies the current medical ultrasound data of the cancer patient.

[0076] Furthermore, in a possible implementation, the classifier is a decision tree; classifying the current medical ultrasound data includes: obtaining various quality indicators in the current medical ultrasound data; determining the maximum value of the comprehensive evaluation value of the dynamic impact of the tumor of the treatment stage to which the current medical ultrasound data belongs, and obtaining the final comprehensive evaluation value of the dynamic impact of the tumor of the current medical ultrasound data; inputting the various quality indicators of the current medical ultrasound data and the final comprehensive evaluation value of the dynamic impact of the tumor into a pre-constructed decision tree, and using the decision tree to classify the current medical ultrasound data to obtain a classification result.

[0077] In a specific example, first, various quality indicators are extracted from the current medical ultrasound data of a tumor patient: Contrast: calculate the gray value difference ratio between the target area and the background; Sharpness: measures the sharpness of image edges (via gradient operator); Noise level: evaluates the standard deviation of grayscale in a uniform area; Artifact Degree: Detects the presence of common artifacts (reverberation / acoustic shadows).

[0078] Secondly, the current medical ultrasound data may belong to multiple treatment stages, so the maximum value of the comprehensive evaluation values ​​of tumor dynamic impact of all treatment stages to which the current medical ultrasound data belongs is determined, and the maximum value is used as the final comprehensive evaluation value of tumor dynamic impact of the current medical ultrasound data.

[0079] Finally, various quality indicators of the current medical ultrasound data and the final comprehensive evaluation value of the dynamic impact of the tumor are input into a pre-built decision tree, and the decision tree is used to classify the current medical ultrasound data to obtain a classification result.

[0080] The pre-construction process of the decision tree includes the following: First, obtaining training sample data: A large amount of medical ultrasound data from patients with the same type of tumor is collected to form the training sample data. Using the same method as described above for obtaining various quality indicators and the final comprehensive evaluation value of the dynamic impact of the tumor in the current medical ultrasound data of the tumor patient, various quality indicators and the final comprehensive evaluation value of the dynamic impact of the tumor are obtained for each medical ultrasound data item in the training sample data. Second, labeling the training sample data: Multiple (e.g., three) experienced ultrasound physicians label each medical ultrasound data item in the training sample data. Then, constructing a decision tree: Using the training sample data and its labeled labels, the decision tree is trained. During the training process, the root node and branches at each level are selected, for example, using the Gini impurity metric to select the most discriminative feature as the root node, thereby completing the construction of the decision tree. Since this embodiment of the present invention builds on existing decision tree construction by adding the comprehensive evaluation value of the dynamic impact of the tumor in the medical ultrasound data as a classification indicator at a certain level in the decision tree, the specific construction process of the decision tree is the same as in the prior art and will not be further described here.

[0081] Figure 2 The diagram shows the structure of the pre-built decision tree. In the root node selection, the Gini impurity index is used, and the feature with the strongest discrimination is the degree of artifacts. When the artifact area accounts for ≤5%, it is directly marked as "unqualified". In the first-level branch, the left branch (artifacts are qualified), the secondary splitting feature is contrast (threshold: target / background grayscale ratio ≥ 2.5), and the right branch is artifacts exceeding the standard, which is directly marked as "unqualified". In the second-level branch, the splitting feature of the branch with qualified contrast is clarity (threshold: Sobel gradient value ≥ 0.15), and the branch with unqualified contrast is marked as "unqualified". In the third-level branch, the splitting feature of the branch with qualified clarity is noise level (threshold: standard deviation of uniform area). ≤25), the branch with unqualified clarity is marked as "boundary correction suggestion" (problem adjustment suggestion, such as using the existing noise reduction algorithm); in the fourth-level branch, the splitting feature of the noise-qualified branch is the tumor dynamic impact comprehensive evaluation value TDII. When TDII ≥ 80, it is marked as "image quality is qualified, but the tumor is in a high activity period, and immediate clinical intervention is recommended". Action: give priority to biopsy / enhanced imaging examination. When 60 ≤ TDII < 80, it is marked as "image quality is qualified, the tumor is moderately active, and it is recommended to shorten the review cycle". Action: re-evaluate ultrasound + hemodynamic monitoring within 45 days. When TDII < 60, it is marked as "image quality is qualified, the tumor is relatively stable, and routine follow-up is recommended". Action: maintain standard follow-up for 6 months.

[0082] The embodiment of the present invention adds a comprehensive evaluation value of the dynamic impact of tumors to the original decision tree, so that the final decision recommendation not only takes into account image quality, but also incorporates factors such as dynamic changes in tumors. This can enhance the classification tree's ability to identify different tumor stages, thereby ultimately improving the classification accuracy of medical ultrasound data.

[0083] Based on the same inventive concept, an embodiment of the present invention also provides a classifier-based medical ultrasound data classification device, which includes: a memory, a processor, and a computer program code stored in the memory and running on the processor, wherein when the processor executes the computer program code, the system can execute any one of the classifier-based medical ultrasound data classification methods introduced above.

[0084] In embodiments of the present invention, the device may be divided into functional modules based on the above-described method examples. For example, these modules may correspond to individual functional modules, or two or more functions may be integrated into a single processing module. The integrated modules may be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and represents only a logical functional division. In actual implementation, other division methods may be employed.

[0085] Based on the same inventive concept, an embodiment of the present invention also provides a classification system for medical ultrasound data based on a classifier, such as Figure 3 As shown, the system includes: A data acquisition module, used to acquire several medical ultrasound data of a tumor patient including the current medical ultrasound data; a stage division module, configured to divide the treatment process of the tumor patient into a plurality of treatment stages based on the detection time of all the medical ultrasound data; an evaluation module, configured to determine a comprehensive evaluation value of the dynamic impact of the tumor during the treatment phase based on a distribution stability of detection time intervals between adjacent medical ultrasound data during the treatment phase and a distribution of different tumor-related indicators in all the medical ultrasound data during the treatment phase; The classification module is used to classify the current medical ultrasound data using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs.

[0086] It should be noted that the system provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above.

[0087] Based on the same inventive concept, an embodiment of the present invention further provides a computer program product, which includes: computer program code, which, when running on a computer, enables the computer to execute any one of the aforementioned classifier-based medical ultrasound data classification methods.

[0088] Based on the same inventive concept, an embodiment of the present invention also provides a computer-readable storage medium, which stores computer program code. When the computer program code runs on a computer, the computer executes any one of the classifier-based medical ultrasound data classification methods introduced above.

[0089] It should be noted that the above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the various embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A classification method for medical ultrasound data based on a classifier, characterized in that: The following steps are involved: Acquiring several medical ultrasound data of a tumor patient including current medical ultrasound data; dividing the treatment process of the tumor patient into several treatment stages based on the detection time of all the medical ultrasound data; Determining a comprehensive evaluation value of the dynamic impact of the tumor during the treatment phase based on a distribution stability of detection time intervals between adjacent medical ultrasound data during the treatment phase and a distribution of different tumor-related indicators in all the medical ultrasound data during the treatment phase; The current medical ultrasound data is classified using a classifier based on the comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage to which the current medical ultrasound data belongs.

2. The method for classifying medical ultrasound data based on a classifier according to claim 1, characterized in that: The treatment process of the tumor patient is divided into several treatment stages, including: determining a time interval sequence of the tumor patient based on time intervals between detection times of adjacent medical ultrasound data; dividing the detection time interval sequence into a plurality of subsequences based on a change trend of a local detection time interval in the detection time interval sequence; A number of treatment stages of the treatment process of the tumor patient are determined based on the number of ultrasound detections corresponding to the detection time intervals in the subsequence.

3. The method for classifying medical ultrasound data based on a classifier according to claim 1, characterized in that: The different tumor-related indicators include at least tumor volume, blood flow resistance index, and calcification point count; determining the comprehensive evaluation value of the dynamic impact of the tumor during the treatment stage includes: determining a stability evaluation coefficient of the treatment stage based on a distribution stability of detection time intervals between adjacent medical ultrasound data in the treatment stage; determining the controllability of tumor manifestation in the treatment phase based on changes in tumor volume and blood flow resistance index in all the medical ultrasound data in the treatment phase; determining the tumor malignant invasion impact degree during the treatment phase based on changes in the number of calcification points and blood flow resistance index in all the medical ultrasound data during the treatment phase; Based on the stability evaluation coefficient, the controllability of the tumor performance and the influence of the tumor malignant invasion are weightedly fused to determine a comprehensive evaluation value of the dynamic impact of the tumor in the treatment stage.

4. The method for classifying medical ultrasound data based on a classifier according to claim 3, characterized in that: Determine the stability assessment coefficient for the treatment phase, including: determining a standard deviation of detection time intervals of all adjacent medical ultrasound data in the treatment phase to obtain a time interval dispersion; A negative correlation normalization process is performed on the time interval dispersion to obtain a stable evaluation coefficient of the treatment stage.

5. The method for classifying medical ultrasound data based on a classifier according to claim 3, characterized in that: Determine the degree of control of tumor manifestations during the treatment phase, including: determining a controllable degree of tumor volume change based on a change rate of the tumor volume in all the medical ultrasound data during the treatment phase; determining the true growth representation of the tumor based on the dispersion of the blood flow resistance index in all the medical ultrasound data during the treatment phase; The actual tumor growth expression degree is used as a weight to perform weighted multiplication and normalization on the tumor volume change controllability to obtain the tumor expression controllability at the treatment stage.

6. The method for classifying medical ultrasound data based on a classifier according to claim 3, characterized in that: Determine the degree of tumor aggressiveness at the treatment stage, including: determining a calcification growth rate expression degree based on the number of calcification points in all the medical ultrasound data and the detection time of all the medical ultrasound data during the treatment stage; determining a steep rise degree of the blood resistance index based on changes in the blood resistance index in all the medical ultrasound data during the treatment phase; The tumor malignant invasion impact degree in the treatment stage is calculated based on the calcification growth rate expression degree and the blood resistance index steepness.

7. The method for classifying medical ultrasound data based on a classifier according to claim 6, characterized in that: Determine the calcification growth rate expression, including: determining a maximum number of calcification points among all the calcification points in the medical ultrasound data during the treatment phase; determining a time interval between the medical ultrasound data corresponding to the maximum number of calcification points and a detection time of the first medical ultrasound data in the treatment stage as a target time interval; Determining the ratio of the target time interval to the entire time interval spanned by the treatment phase as the maximum calcification penetration; Normalizing the ratio of the maximum number of calcification points to the number of calcification points in the medical ultrasound data detected by the first ultrasound examination during the treatment phase to obtain a maximum calcification expression; The maximum calcification expression penetration is multiplied by the maximum calcification expression to obtain the calcification growth rate expression.

8. The method for classifying medical ultrasound data based on a classifier according to claim 6, characterized in that: Determine the steepness of the blood resistance index, including: determining target medical ultrasound data corresponding to a maximum number of calcification points among all the medical ultrasound data in the treatment stage; determining a maximum ratio of a blood resistance index in a subsequent medical ultrasound data of the target medical ultrasound data to a blood resistance index in each preceding medical ultrasound data; The maximum ratio is normalized to obtain the steep rise degree of the blood resistance index.

9. The method for classifying medical ultrasound data based on a classifier according to claim 1, characterized in that: The classifier is a decision tree; Classifying the current medical ultrasound data includes: Obtaining various quality indicators in the current medical ultrasound data; Determining a maximum value among the comprehensive evaluation values ​​of dynamic tumor impacts of the treatment stage to which the current medical ultrasound data belongs, and obtaining a final comprehensive evaluation value of dynamic tumor impacts of the current medical ultrasound data; Various quality indicators of the current medical ultrasound data and the final comprehensive evaluation value of the dynamic impact of the tumor are input into a pre-built decision tree, and the current medical ultrasound data is classified using the decision tree to obtain a classification result.

10. A classification device for medical ultrasound data based on a classifier, characterized in that: The method comprises a memory, a processor, and an executable computer program code stored in the memory and runnable on the processor, wherein when the processor executes the computer program code, the method for classifying medical ultrasound data based on a classifier as described in any one of claims 1 to 9 is executed.

Images