Prostate image-based assessment method, apparatus, device, medium, and product

Abstract

The present disclosure relates to a prostate image-based evaluation method, device, equipment, medium and product, the method comprising: acquiring first image data of a prostate target area matched with a structure evaluation model, the structure evaluation model comprising a structure morphology network and a structure evaluation network; performing feature extraction on the first image data through the structure morphology network to obtain morphology feature data, the morphology feature data representing quantitative information of a structure morphology of the prostate target area; and performing structure evaluation according to the morphology feature data through the structure evaluation network to determine structure evaluation data of the prostate target area. The present embodiment solves the problem of insufficient accuracy of an artificial evaluation mode, guarantees the objectivity and repeatability of prostate evaluation tasks, and provides assistance for promoting the standardization iteration of prostate intervention technology.

Description

Technical Field

[0001] This disclosure relates to the field of medical image processing technology, and in particular to an evaluation method, apparatus, device, medium and product based on prostate images. Background Technology

[0002] Accurate assessment of prostate function is a core decision-making basis for optimizing intervention programs and achieving personalized interventions, playing a crucial role in improving intervention effectiveness and reducing adverse risks. Current assessment models primarily rely on manual observation of anatomical characteristics of key glandular structures in the prostate, such as size, surface texture, firmness, and capsule integrity, combined with subjective judgment based on clinical experience.

[0003] The aforementioned assessment model is greatly influenced by subjective factors, and due to the lack of a unified quantitative anchoring standard, it is impossible to construct a universal and standardized assessment process, and it is also difficult to ensure the objectivity and repeatability of the assessment task, thus restricting the standardized iteration of prostate intervention technology. Summary of the Invention

[0004] This disclosure provides a prostate image-based assessment method, apparatus, device, medium, and product to address the inaccuracy of manual assessment methods, achieve objectivity and repeatability in prostate assessment tasks, and contribute to the standardized iteration of prostate intervention technologies.

[0005] One aspect of this disclosure provides an assessment method based on prostate images, the method comprising:

[0006] First image data of the prostate target region matching the structural evaluation model, which includes a structural morphology network and a structural evaluation network, are acquired.

[0007] The first image data is subjected to feature extraction through the morphological network to obtain morphological feature data, which represents the quantitative information of the structural morphology of the prostate target area.

[0008] The structural evaluation network performs structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target area.

[0009] Another aspect of this disclosure provides an assessment apparatus based on prostate images, the apparatus comprising:

[0010] The first image data acquisition module is used to acquire first image data of the prostate target area that matches the structural evaluation model, wherein the structural evaluation model includes a structural morphology network and a structural evaluation network.

[0011] The morphological feature data determination module is used to extract features from the first image data through the structural morphology network to obtain morphological feature data, wherein the morphological feature data represents the quantitative information of the structural morphology of the prostate target area;

[0012] The structural assessment data determination module is used to perform structural assessment based on the morphological feature data through the structural assessment network to determine the structural assessment data of the prostate target area.

[0013] Another aspect of this disclosure provides an electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the prostate image-based evaluation method according to any embodiment of this disclosure.

[0017] Another aspect of this disclosure provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the prostate image-based evaluation method according to any embodiment of this disclosure.

[0018] Another aspect of this disclosure provides a computer program product including a computer program that, when executed by a processor, implements the prostate image-based evaluation method described in any embodiment of this disclosure.

[0019] The technical solution of this disclosure constructs a structural assessment model by planning and quantifying features from the structural morphology dimension of the prostate. Through the structural morphology network in the structural assessment model, features are extracted from the first image data of the prostate target area to obtain morphological feature data. The structural assessment network in the structural assessment model then evaluates the morphological feature data to determine the structural assessment data of the prostate target area. By utilizing the standardized quantitative representation of the prostate's structural morphology in imaging, the objectivity and repeatability of the prostate assessment task are ensured. This solves the problem of insufficient accuracy in assessment models that rely on manual methods, providing precise data support for the analysis and judgment of prostate function, and thus contributing to the standardized iteration of prostate intervention technology.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart illustrating a prostate image-based evaluation method provided in one embodiment of this disclosure;

[0023] Figure 2 A flowchart illustrating another prostate image-based evaluation method provided in one embodiment of this disclosure;

[0024] Figure 3 A flowchart illustrating another prostate image-based evaluation method provided in one embodiment of this disclosure;

[0025] Figure 4 This is a schematic diagram of the structure of a prostate image-based evaluation device provided in one embodiment of the present disclosure;

[0026] Figure 5 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present disclosure. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present disclosure, the technical solutions of the present disclosure will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present disclosure, and not all embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present disclosure.

[0028] It should be noted that the terms "first," "second," "third," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Figure 1 This is a flowchart illustrating a prostate image-based assessment method according to an embodiment of this disclosure. This embodiment is applicable to situations where prostate structure is assessed based on image data. The method can be executed by a prostate image-based assessment device, which can be implemented in hardware and / or software and can be configured in a terminal device. Figure 1 As shown, the method includes:

[0030] S110. Acquire first image data of the prostate target area that matches the structural evaluation model.

[0031] Specifically, the structural evaluation model is used to quantitatively evaluate the structural attributes of the prostate target area. In this embodiment, the structural evaluation model includes a structural morphology network and a structural evaluation network. The structural morphology network extracts quantitative features of the structural morphology of the prostate target area from the input image data. Its core function is to convert image information into a quantitative representation that can be understood and processed by the structural evaluation network. The structural evaluation network quantifies the structural attributes of the prostate target area based on the output of the structural morphology network. Its core function is to complete higher-level attribute determination based on the quantified morphology. Together, they achieve end-to-end processing from prostate imaging to structural evaluation.

[0032] The structure and morphology of the prostate are highly coupled with its physiological functions such as secretion, urination, and reproduction. Abnormal changes in its structure and morphology are often accompanied by damage to the physiological functions of the prostate. For example, atrophy of the glandular tissue or reduction of acini in the prostate is often accompanied by abnormal secretory function. An enlarged prostate can cause urinary dysfunction such as difficulty urinating and incomplete urination. Abnormal morphologies such as dilation, tortuosity, or obstruction of the ducts in the prostate can affect semen parameters, such as sperm motility, thereby causing damage to reproductive function.

[0033] In this embodiment, the structural assessment model describes the purpose of evaluating the structural attributes of the prostate based on image data of the prostate structure. It aims to transform the structural assessment requirements into an evidence-based, systematic, and repeatable analysis and judgment process, thereby providing data support for the analysis and judgment of prostate function.

[0034] The prostate target area refers to the region of interest in the prostate for structural assessment, encompassing one or more anatomical regions. These anatomical regions contain complete or partial morphological information of one or more prostate structures. For example, in terms of physiological function, prostate structures associated with secretory function may be prostate secretory ducts; exemplarily, the prostate target area may include the transitional zone, central zone, and peripheral zone. Prostate structures associated with reproductive function may be the verumontanum; exemplarily, the prostate target area includes the verumontanum. Prostate structures associated with urination function may be the prostate urethra; exemplarily, the prostate target area includes one or more longitudinal anatomical regions. These anatomical structures and regions are provided as examples only and are not intended to be limiting.

[0035] For example, the prostate target area can be a segmented region composed of multiple anatomical regions, where one anatomical region encompasses one or more anatomical structures. Alternatively, it can be a continuous region composed of multiple anatomical regions, including not only the anatomical regions defined by the anatomical structures but also the connecting regions between these regions. Of course, the prostate target area can also be the entire prostate. In this embodiment, the prostate target area can be pre-defined based on the structural assessment requirements expressed by the structural assessment model.

[0036] In an optional embodiment, when the structural assessment model covers the structural assessment needs of the prostatic urethra, the prostatic target area includes at least one longitudinal anatomical region among the bladder neck, the middle of the prostate, and the apex of the prostate. The bladder neck, the middle of the prostate, and the apex of the prostate are anatomical segments directly spatially associated with the prostatic urethra, and each longitudinal anatomical region contains important morphological information about the prostatic urethra, whose morphological characteristics directly affect the patency of the prostate.

[0037] Specifically, the bladder neck is the initial region where the bladder transitions to the prostatic urethra. The morphology of the prostatic urethra at the bladder neck directly reflects the initial patency of urine flowing from the bladder into the prostatic urethra, making it an important area for assessing upstream voiding function. The middle portion of the prostate, located between the bladder neck and the apex, is a common site for abnormal tissue formation. The morphology of the prostatic urethra in the middle portion is related to the hydrodynamic efficiency of the prostatic urethra, directly reflecting the mid-stream patency of urine flowing through the prostatic urethra, such as staged attenuation or abnormal fluctuations. The apex of the prostate is the final region where the prostatic urethra transitions to the membranous urethra, densely populated with the peripheral sphincter complex. The morphology of the prostatic urethra at the apex directly reflects the final patency of urine flowing from the prostatic urethra into the membranous urethra, making it a key gateway for assessing downstream voiding function.

[0038] In this embodiment, the first image data represents a set of image information obtained through image information processing, with the prostate target area as the core. The first image data consists of at least two original image frames output by the image acquisition device, and the image information processing includes two-dimensional registration processing and / or three-dimensional reconstruction processing. For example, the first image data can be obtained by filtering from the entire prostate's global image data based on the target area defined by the prostate target area, or it can be specifically acquired in the prostate target area using an image acquisition device. The image acquisition device can be an ultrasound diagnostic instrument, magnetic resonance imaging device, computed tomography (CT) scanner, positron emission tomography (PET) scanner, or cystoscope, etc., but is not limited to the examples given above.

[0039] In the ablation procedure, the structural assessment model covers the structural assessment needs of the prostatic urethra. In one optional embodiment, the method further includes controlling the cystoscope to acquire images during the cystoscopy exit phase after the ablation procedure. Specifically, the movement of the cystoscope and image acquisition can be automated by a controller or manually performed by the operator.

[0040] In an alternative embodiment, the prostate target area is a segmented region, and accordingly, controlling the cystoscope to acquire images includes: in response to the cystoscope passing through a longitudinal anatomical region, controlling the cystoscope to acquire images in the longitudinal anatomical region.

[0041] In a specific example, after the ablation operation is completed, the cystoscope is withdrawn from the bladder cavity toward the external urethral orifice. For example, image acquisition is performed within a region defined by the bladder neck when the field of view first fully reveals the V-shaped or trumpet-shaped structure of the internal urethral orifice, or when the cumulative travel of the cystoscope reaches a first travel threshold. The cystoscope is then withdrawn further toward the external urethral orifice. For example, image acquisition is performed within a region defined by the middle of the prostate when the tissue pressure reported by the pressure sensor exceeds a pressure threshold, the field of view exhibits axial symmetry, the structural deformation presented in the field of view conforms to a predefined compression deformation, or the cumulative travel of the cystoscope reaches a second travel threshold. The cystoscope is then withdrawn further toward the external urethral orifice. For example, image acquisition is performed within a region defined by the apex of the prostate when the urethral mucosa in the field of view exhibits abrupt changes in the shape of annular folds, the urethral lumen diameter exhibits a staged narrowing characteristic, the pixel ratio of the verumontanum in the field of view meets a preset ratio range, or the cumulative travel of the cystoscope reaches a third travel threshold. The first, second, and third travel thresholds increase sequentially.

[0042] The above embodiments are merely illustrative examples of the location of longitudinal anatomical regions, but are not limited to the examples given above.

[0043] In another alternative embodiment, the prostate target area is a continuous region. Accordingly, controlling the cystoscope to perform image acquisition includes: in response to the cystoscope passing through the first longitudinal anatomical region, controlling the cystoscope to start the image acquisition task, and controlling the cystoscope to end the image acquisition task after completing the image acquisition of the last longitudinal anatomical region.

[0044] In another alternative embodiment, the prostate target area is the entire prostate, and the control of the cystoscope to perform image acquisition includes: in response to receiving a cystoscope exit command, controlling the cystoscope to start an image acquisition task, and until receiving an exit completion command, controlling the cystoscope to end the image acquisition task.

[0045] Based on the above embodiments, exemplarily, during image acquisition, in response to the imaging quality not meeting the quality conditions, the cystoscope's shooting pose and / or shooting parameters are adjusted until the imaging quality meets the quality conditions, at which point the acquisition of original image frames begins or the original image frames are re-acquired. Exemplarily, imaging quality can be represented by quality parameters such as real-time sharpness, edge clarity, color reproduction, brightness uniformity, and noise level, but is not limited to the given examples.

[0046] Based on the above embodiments, optionally, the first image data is a two-dimensional image sequence, a three-dimensional reconstruction model, mixed image data, or multi-dimensional image data.

[0047] Specifically, mixed image data indicates that at least one anatomical region uses an imaging mode different from other anatomical regions, while multidimensional image data indicates that at least one anatomical region uses two imaging modes simultaneously.

[0048] For example, suppose the prostate target area includes anatomical region A and anatomical region B. Both anatomical regions can be imaged in two-dimensional mode, and a two-dimensional image sequence can be obtained after two-dimensional registration. Alternatively, both anatomical regions can be imaged in three-dimensional mode, and a three-dimensional reconstruction model can be obtained after three-dimensional reconstruction. Anatomical region A can be imaged in two-dimensional mode, and anatomical region B can be imaged in three-dimensional mode. After image information processing, mixed image data can be obtained. Both anatomical regions can be imaged in two-dimensional and three-dimensional modes simultaneously. Of course, anatomical region A can be imaged in two-dimensional and three-dimensional modes simultaneously, while anatomical region B can be imaged in two-dimensional mode. After image information processing, multi-dimensional image data can be obtained.

[0049] The imaging mode corresponding to the first image data is not limited here. It is understood that the embodiments of this disclosure support prostate assessment tasks in both two-dimensional imaging mode and three-dimensional imaging mode.

[0050] In one specific embodiment, the structural assessment model covers the structural assessment needs of the prostatic urethra. The prostatic target area includes the entire prostate or at least one longitudinal anatomical region. The two-dimensional image sequence in the first image data consists of multiple registered cross-sectional images. For example, the two-dimensional image sequence corresponding to the longitudinal anatomical region includes 2-4 cross-sectional images within the area defined by the longitudinal anatomical region, which has the advantages of fast imaging speed, low equipment cost, and strong real-time performance. The image data corresponding to the three-dimensional reconstruction model in the first image data consists of a continuous multi-angle video sequence to ensure sufficient parallax coverage and reconstruction redundancy between frames. For example, the continuous multi-angle represents the micro-rotation angles that progress successively within the range of 10°–15° along the exit trajectory of the cystoscope.

[0051] S120. The first image data is extracted using the structural morphology network to obtain morphological feature data.

[0052] In this embodiment, the morphological feature data represents the quantitative information of the structural morphology of the prostate target area, and the morphological feature data includes at least one structural morphological feature corresponding to each prostate structure.

[0053] In an alternative embodiment, the structural morphological features simultaneously satisfy the following conditions: 1) they have a physical definition and can be extracted from image data; 2) they are related to the physiological function of the prostate; and 3) they are repeatable and operable.

[0054] Condition 1) is used to avoid the impact of vaguely defined structural morphological features on the accuracy of prostate structure assessment. For example, glandular vitality is a functional description and cannot be extracted from image data through image processing technology; it usually relies on other detection methods for inference. Condition 2) is used to ensure the practicality of prostate structure assessment and avoid purely theoretical indicators that are detached from practical application value. For example, the ellipticity deviation of the prostate contour currently lacks sufficient clinical research to prove that it has a stable and explainable correlation with the physiological function of the prostate. Condition 3) repeatability means that under the same or similar conditions, the structural morphological features are consistent across different operators, different devices, or different times, to avoid random errors causing excessive fluctuations in prostate structure assessment and affecting its reliability. For example, the complexity of glandular texture is greatly affected by the ambiguity of the scoring criteria. Operability means that the extraction process of structural morphological features is adapted to the actual application scenario and meets the conventional operational requirements such as convenient operation, reasonable time consumption, and universal equipment. For example, fine gaps are highly dependent on the imaging resolution of the image acquisition equipment rather than the anatomical structure.

[0055] In one optional embodiment, the structural morphology network includes a morphological feature layer corresponding to each structural morphological feature. The morphological feature layer integrates image processing algorithms to perform targeted feature extraction on the first image data, such as edge detection, region segmentation, texture analysis, morphological measurement, and deep learning networks, and outputs structural morphological features.

[0056] S130. The structural evaluation network is used to perform structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target area.

[0057] Specifically, structural assessment data represents a standardized, interpretable, and structured set of assessment results describing the structural morphology of the prostate target area, including but not limited to quantitative scoring, grading, status description, anomaly alerts, and risk warnings. For example, structural assessment data can be applied to technical scenarios related to prostate intervention techniques, providing data support for pre-intervention planning, assessment feedback during intervention, immediate post-intervention assessment, long-term risk prediction, and protocol optimization. Interventions can include ablation, medication administration, and thermotherapy, but are not limited to the examples given above.

[0058] In one optional embodiment, the structural evaluation data includes the structural evaluation results of the prostate target region, wherein the structural evaluation results include a first evaluation result corresponding to the prostate target region and / or a second evaluation result corresponding to each structural morphological feature. The first evaluation result represents the structural evaluation information of the prostate structure in a global dimension, and the second evaluation result represents the structural evaluation information of the prostate structure in a local dimension.

[0059] In one optional embodiment, the first evaluation result includes at least one of a first evaluation score, a first evaluation grade, and a first evaluation conclusion, and the second evaluation result includes at least one of a second evaluation score, a second evaluation grade, and a second evaluation conclusion.

[0060] Among them, the evaluation score represents the parameter value obtained by continuous evaluation of the prostate structure, which precisely reflects the quality of the prostate structure; the evaluation level represents the parameter value obtained by discrete evaluation of the prostate structure, which is used to classify the prostate structure qualitatively or semi-quantitatively; and the evaluation conclusion is the parameter conclusion obtained by state determination of structural attributes, which is used to describe the morphological performance, quality determination and other information of the prostate structure in a global or local dimension.

[0061] In an optional embodiment, the structural evaluation network includes a structural evaluation layer corresponding to each structural morphological feature. The structural evaluation network performs structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target region, including: inputting each structural morphological feature into the corresponding structural evaluation layer to obtain the respective output second evaluation result.

[0062] For example, the structural evaluation network is a normalization layer or an activation layer. The activation layer is used to determine the reasonableness of structural morphological features. For example, the activation layer integrates normal value ranges, abnormal value ranges, and normal or abnormal feature thresholds. The normal value range represents the acceptable numerical fluctuation range of the structural morphological features. The second evaluation result determined based on the normal value range or the normal feature threshold can be used to reflect the risk level of abnormal structures within the prostate target area. The abnormal value range represents the unacceptable numerical fluctuation range of the structural morphological features. The second evaluation result determined based on the abnormal value range or the abnormal feature threshold can be used to reflect the tendency for abnormal structures to appear within the prostate target area.

[0063] Taking the normal value range of the activation layer integration as an example, when the structural morphology feature is within the normal value range, the offset within the interval is determined based on the median of the normal value range, and the second evaluation result corresponding to the structural morphology feature is determined based on the offset within the interval; when the structural morphology feature is not within the normal value range, the offset outside the interval is determined based on the upper or lower limit of the normal value range, and the second evaluation result corresponding to the structural morphology feature is determined based on the offset outside the interval.

[0064] In another optional embodiment, the structural evaluation network includes a normalization layer and a feature aggregation layer. The step of performing structural evaluation based on the morphological feature data through the structural evaluation network to determine the structural evaluation data of the prostate target region includes: inputting the morphological feature data into the normalization layer to obtain the output normalized feature data; and performing aggregation processing on the normalized feature data through the feature aggregation layer to obtain a first evaluation result of the prostate target region.

[0065] The normalization layer contains one or more normalization algorithms. The purpose of normalization is to eliminate the dimensional differences of different structural features, which can be adapted to prostate assessment scenarios across samples. It avoids interference caused by individual physiological variations and differences in image acquisition equipment, and ensures the adaptability and stability of the prostate assessment task.

[0066] In the above embodiments, the normalization algorithms, exemplarily, include but are not limited to piecewise normalization, Min-Max normalization, or Z-score standardization. Piecewise normalization refers to analyzing the distribution characteristics of structural morphological features using a certain amount of sample set, defining piecewise feature mappings corresponding to the structural morphological features based on their anatomical significance, and accurately mapping the structural morphological features to [0,1]. Min-Max normalization involves extracting the maximum and minimum values ​​of the structural morphological features in the sample set, and constraining the structural morphological features within the [0,1] interval based on these values. Z-score standardization refers to standardizing the structural morphological features for approximately normally distributed structural morphological features based on their mean and standard deviation in the sample set.

[0067] Specifically, the normalization algorithms corresponding to different structural morphological features can be the same or different. Based on the distribution characteristics of the structural morphological features, the matching normalization algorithm in the normalization layer can be obtained to ensure that different structural morphological features have a relatively balanced influence, avoiding the dominance of the first evaluation result by a single structural morphological feature due to differences in dimensions or value range. As the sample set accumulates, the normalization layer in the structural evaluation network can be iteratively optimized based on new samples to continuously improve the stability and adaptability of the normalization process, thereby ensuring the accuracy of prostate structure evaluation.

[0068] The feature aggregation layer is an integrated analysis model that uses standardized aggregation algorithms to transform morphological feature data into structural attribute information at the global dimension. For example, the feature aggregation layer can be a logical decision tree, a weighted fusion layer, a machine learning model, or a deep learning network, but it is not limited to the given examples. For instance, when the feature aggregation layer is a weighted fusion layer, it defines feature weight data, which is a set of numerical coefficients used to quantify the relative contribution or relative importance of different structural morphological features.

[0069] In the above embodiments, the first evaluation result represents the absolute evaluation information of the prostate structure, reflecting the overall quality of the prostate structure.

[0070] In another optional embodiment, the structural evaluation network includes a feature aggregation layer and an activation layer corresponding to each structural morphological feature. The step of performing structural evaluation based on the morphological feature data through the structural evaluation network to determine the structural evaluation data of the prostate target region includes: determining a second evaluation result corresponding to each structural morphological feature based on the morphological feature data through each activation layer; and inputting each second evaluation result into the feature aggregation layer for aggregation processing to obtain a first evaluation result of the prostate target region.

[0071] In this embodiment, the first assessment result represents the relative assessment information of the prostate structure, reflecting its overall deviation from the normal morphology or its overall tendency towards the abnormal morphology.

[0072] The technical solution of this embodiment constructs a structural assessment model by planning and quantifying features from the structural morphology dimension of the prostate. Through the structural morphology network in the structural assessment model, features are extracted from the first image data of the prostate target area to obtain morphological feature data. The structural assessment network in the structural assessment model evaluates the morphological feature data to determine the structural assessment data of the prostate target area. By utilizing the standardized quantitative representation of the prostate's structural morphology in imaging, the objectivity and repeatability of the prostate assessment task are ensured, solving the problem of insufficient accuracy of assessment models that rely on manual methods. This provides accurate data support for the analysis and judgment of prostate function, and further contributes to promoting the standardized iteration of prostate intervention technology.

[0073] Figure 2 This is a flowchart of another prostate image-based evaluation method provided in one embodiment of this disclosure. This embodiment further refines the step of "extracting features from the first image data using the structural morphology network to obtain morphological feature data" in the above embodiment. In this embodiment, the step of extracting features from the first image data using the structural morphology network to obtain morphological feature data includes: determining third image data corresponding to each morphological feature layer based on the first image data using the image input layer; extracting features from the input third image data using each morphological feature layer to obtain the structural morphological features of the prostatic urethra; and determining morphological feature data based on each of the structural morphological features. Figure 2 As shown, the method includes:

[0074] S210. Acquire first image data of the prostate target area that matches the structural evaluation model.

[0075] In an optional embodiment, the structural evaluation model includes a first evaluation model and / or a second evaluation model, wherein the first evaluation model is used to evaluate the structural properties of the prostate target area in two-dimensional imaging mode and / or three-dimensional imaging mode, and the second evaluation model is used to evaluate the structural properties of the prostate urethra within the prostate target area in one or more patency characteristic dimensions.

[0076] Specifically, the first evaluation model can run independently to meet the structural evaluation requirements under the imaging mode dimension, and the second evaluation model can run independently to meet the structural evaluation requirements under the patency characteristic dimension. The two can also run together to meet the structural evaluation requirements under the hierarchical evaluation dimension, such as evaluating the performance of the structural morphology of the prostatic urethra related to radial patency characteristics in the two-dimensional imaging mode.

[0077] The first evaluation model focuses on the imaging patterns of the prostate structure, comprehensively evaluating its morphological appearance in two-dimensional and / or three-dimensional imaging modes. Specifically, the morphological feature layer in the two-dimensional evaluation model extracts the structural morphological features of the prostate structure in two-dimensional image sequences, and its output structural morphological features possess two-dimensional image attributes, such as the angle, length, geometric thickness, and area of ​​the prostate structure in two-dimensional imaging mode. The morphological feature layer in the three-dimensional evaluation model extracts the structural morphological features of the prostate structure in a three-dimensional reconstruction model, and its output structural morphological features possess three-dimensional image attributes, such as the angle, length, area, and volume of the prostate structure in three-dimensional imaging mode.

[0078] The second assessment model focuses on the patency characteristics of the prostate structure, and is used to evaluate the morphological performance of the prostatic urethra across one or more patency characteristic dimensions. For example, the patency characteristic dimensions can be the initial patency dimension, the middle patency dimension, or the terminal patency dimension, etc., but are not limited to the example scenario.

[0079] In one optional embodiment, the second evaluation model includes at least one of a first patency model, a second patency model, and a third patency model. Specifically, the first patency model is used to evaluate the structural properties of the prostatic urethra in the radial patency dimension, the second patency model is used to evaluate the structural properties of the prostatic urethra in the axial patency dimension, and the third patency model is used to evaluate the structural properties of the prostatic urethra in the spatial patency dimension.

[0080] Based on the above embodiments, optionally, when the structural evaluation model includes a first unobstructed model, the structural morphology network includes a geometric feature layer corresponding to the first anatomical region; when the structural evaluation model includes a second unobstructed model, the structural morphology network includes a symmetry feature layer and / or a coherence feature layer; and when the structural evaluation model includes a third unobstructed model, the structural morphology network includes an elimination feature layer. Correspondingly, when the structural evaluation model includes a first evaluation model, the structural morphology network includes a symmetry feature layer and / or a geometric feature layer corresponding to the first anatomical region; and when the first evaluation model includes a three-dimensional evaluation model, the structural morphology network further includes a coherence feature layer and / or an elimination feature layer.

[0081] In this embodiment, the structural assessment model only covers the structural assessment needs of the prostatic urethra. The first anatomical region includes the bladder neck and / or the apex of the prostate in the prostate target area. The urethral geometric features output by the geometric feature layer represent the quantitative information of the opening morphology of the prostatic urethra in the first anatomical region.

[0082] The geometric feature layer is used to extract quantitative features of the radial morphology of the prostatic urethra. The urethral geometric features corresponding to the bladder neck are used to quantify the degree of opening of the prostatic urethra at the bladder neck, primarily referring to the opening degree of the internal urethral orifice. A larger bladder neck opening angle indicates a more open internal urethral orifice. The bladder neck opening area is positively correlated with the urine flow rate, directly determining the initial flow rate of urine through the prostatic urethra. The urethral geometric features corresponding to the prostatic apex are used to quantify the degree of opening of the prostatic urethra at the prostatic apex, primarily referring to the opening degree of the external urethral orifice. The opening area at the prostatic apex directly determines the final flow rate of urine through the prostatic urethra.

[0083] The symmetry feature layer and the coherence feature layer are used to extract quantitative features of the axial morphology of the prostatic urethra. The urethral symmetry features output by the symmetry feature layer characterize the relative symmetry, rather than absolute symmetry, of the two urethral halves obtained by axially dividing the prostatic urethra. Relative symmetry refers to whether the morphological appearance of the two urethral halves tends to be consistent or whether the deviation is within a reasonable threshold range. Absolute symmetry refers to whether the two urethral halves are completely consistent or without deviation; typically, the two urethral halves are distributed anteriorly or posteriorly or laterally along the urethral axis. Specifically, urethral symmetry features can be used to quantitatively describe the axial symmetry of the prostatic urethra in each longitudinal anatomical region, and also to quantitatively describe the axial symmetry of the prostatic urethra throughout the entire prostatic target area. Low urethral symmetry features indicate uneven urine flow distribution, which increases the risk of urethral stricture, making it particularly suitable for immediate operational assessment after ablation procedures.

[0084] The continuity feature layer outputs urethral continuity features, which characterize the morphological continuity of the prostatic urethra along its axis. This indicates whether the prostatic urethra is uniform, intact, without obvious breaks, or has collapsed, providing data support for damage analysis or risk assessment of voiding function. The elimination feature layer outputs tissue elimination features, representing the volume changes of the prostatic urethra within the prostatic target area after intervention, compared to before the intervention. This can be used to assess the elimination of abnormal tissue in the prostatic urethra, providing data support for the analysis and assessment of voiding function recovery.

[0085] S220. Based on the first image data, the image input layer determines the third image data corresponding to each morphological feature layer.

[0086] In this embodiment, the morphological feature layer is the geometric feature layer, the symmetry feature layer, the coherence feature layer, or the elimination feature layer. The third image data corresponding to the geometric feature layer and the symmetry feature layer is a two-dimensional image sequence composed of one or more cross-sectional images. The third image data corresponding to the coherence feature layer and the elimination feature layer is a three-dimensional reconstruction model of the prostate target area.

[0087] In embodiments where the morphological feature layer is a geometric feature layer or a symmetry feature layer, optionally, the image input layer determines the third image data corresponding to each morphological feature layer based on the first image data, including: filtering image data in the first image data that matches the morphological feature layer; when the filtered image data is a two-dimensional image sequence, the filtered image data is used as the third image data corresponding to the morphological feature layer. When the filtered image data is a three-dimensional reconstruction model, the filtered image data is equidistantly segmented along the transverse plane to obtain the third image data corresponding to the morphological feature layer.

[0088] Specifically, if the selected image data is a two-dimensional image sequence, the structural morphological features output by the morphological feature layer have two-dimensional image attributes; if the selected image data is a three-dimensional reconstruction model, the structural morphological features output by the morphological feature layer have three-dimensional image attributes.

[0089] S230. By extracting features from the input third image data through each morphological feature layer, the structural morphological features of the prostatic urethra are obtained.

[0090] In one optional embodiment, the morphological feature layer is a geometric feature layer, which includes an angle extraction module and / or an area extraction module, and the third image data is image data corresponding to the bladder neck or the tip of the prostate.

[0091] In this embodiment, the step of extracting features from the input third image data for each morphological feature layer to obtain the structural morphological features of the prostatic urethra includes: determining at least one pair of first urethral contours corresponding to the prostatic urethra based on the third image data using the angle extraction module, and determining the opening angle corresponding to the first anatomical region based on each pair of first urethral contours, wherein the two first urethral contours in the first urethral contour pair are distributed on opposite sides along the urethral axis; and / or determining at least one urethral closure contour corresponding to the prostatic urethra based on the third image data using the area extraction module, and determining the opening area corresponding to the first anatomical region based on each urethral closure contour.

[0092] Taking the bladder neck as an example, in the embodiment of the angle extraction module, the urethral axis within the bladder neck is used as the axis of symmetry. The prostatic urethra within the bladder neck is axially divided along a preset angle in the cross-sectional plane to obtain two first urethral halves. The first urethral contour represents the boundary information of the first urethral halves in the cross-sectional image. The preset angle satisfies the value range of [0, 90]. Taking 0 degrees corresponding to the left and right directions of the imaging object as an example, typically, when the preset angle is 0 degrees, the two first urethral halves are the left urethral halves and the right urethral halves; when the preset angle is 90 degrees, the two first urethral halves are the ventral urethral halves and the dorsal urethral halves.

[0093] For example, for each first urethral contour, the set of boundary points corresponding to the first urethral contour is extracted, and the gradient calculation or least squares fitting is performed on the set of boundary points to obtain the tangent vector corresponding to the first urethral contour; the angle difference between the two corresponding tangent vectors of the first urethral contour is extracted, and the statistical value corresponding to each angle difference is used as the bladder neck opening angle. The statistical value can be the maximum value, minimum value, or average value, etc., but is not limited to the given example.

[0094] In an embodiment of the area extraction module, the inner wall of the prostatic urethra within the bladder neck is used as the contour boundary. Pixels located within the contour boundary in the cross-sectional image are selected, and the connected region formed by the selected pixels is used as the urethral closure contour.

[0095] For example, the number of pixels within each urethral closure contour is counted, and the statistical value corresponding to each pixel count is used as the bladder neck opening area; or, polygon integration is performed on each urethral closure contour to obtain an area estimate, and the statistical value corresponding to each area estimate is used as the bladder neck opening area. The statistical value can be a maximum, minimum, or average value, but is not limited to the given example.

[0096] In another optional embodiment, the morphological feature layer is a symmetrical feature layer, and the third image data includes image data corresponding to at least one second anatomical region, wherein the second anatomical region is the bladder neck, middle part of the prostate, apex of the prostate, or the entire prostate target area in the prostate target area.

[0097] In this embodiment, the step of extracting features from the input third image data through each morphological feature layer to obtain the structural morphological features of the prostatic urethra includes: obtaining second image data matching the second anatomical region from the third image data for each second anatomical region through the symmetry feature layer; determining the longitudinal central axis of the prostatic urethra within the second anatomical region based on the second image data; and performing a mirror comparison operation based on the longitudinal central axis and the second image data to obtain the urethral symmetry features corresponding to the second anatomical region.

[0098] Specifically, the cross-sectional image specified in the second image data is identified to obtain the urethral center point. Within the area defined by the second anatomical region, a longitudinal line segment perpendicular to the cross-sectional plane and passing through the urethral center point is taken as the longitudinal central axis. For example, the specified cross-sectional image can be the cross-sectional image corresponding to the largest urethral area. Using the longitudinal central axis as the axis of symmetry, the prostatic urethra within the second anatomical region is longitudinally divided along a preset angle within the cross-sectional plane to obtain two second urethral halves. The second urethral contour pairs of the two second urethral halves are identified in each cross-sectional image.

[0099] The symmetry feature data represents the degree of asymmetry in the axial morphology of the prostatic urethra within the second anatomical region. The symmetry feature data includes a sequence of asymmetric features corresponding to each cross-sectional image in the second image data. For example, the asymmetric feature sequence includes, but is not limited to, area asymmetry, centroid asymmetry, or distance asymmetry. For each pair of second urethral contours, the contour area corresponding to each second urethral contour in the pair is extracted, and the area ratio of the two contour areas is calculated. The statistical value of each area ratio is used as the area asymmetry. For each pair of second urethral contours, the geometric centroid corresponding to each second urethral contour is extracted, and the longitudinal central axis is used as the mirror axis to determine the mirror centroid corresponding to one of the geometric centroids. The pixel distance between the mirror centroid and the other geometric centroid is calculated. Based on the pixel distance and the inner diameter length of the second urethral contour corresponding to the geometric centroid, the centroid offset feature value is determined, and the statistical value of each centroid offset feature value is used as the centroid asymmetry. For each pair of second urethral contours, one of the second urethral contours is mirrored to obtain a mirror urethral contour. The two maximum pixel distances are calculated by bidirectional directed Hausdorff distance between the mirror urethral contour and the other second urethral contour. The average value corresponding to the two maximum pixel distances is used as the distance offset, and the statistical value of each distance offset is used as the distance asymmetry. The statistical value can be a maximum value, minimum value, or average value, but is not limited to the given example.

[0100] For example, the urethral symmetry value is obtained by performing reverse aggregation calculation on each asymmetric feature sequence, and the statistical value corresponding to multiple urethral symmetry values ​​is used as the urethral symmetry feature. The statistical value can be the maximum value, average value or median value, etc., but is not limited to the given example.

[0101] In another optional embodiment, the morphological feature layer is a coherent feature layer, and the third image data is a three-dimensional reconstruction model of the prostate target area. The step of extracting features from the input third image data through each morphological feature layer to obtain the structural morphological features of the prostate urethra includes: determining a continuous sequence of cross-sectional images based on the third image data through the coherent feature layer; determining axial offset data based on every two adjacent cross-sectional images in the cross-sectional image sequence; and determining the coherent features of the urethra corresponding to the prostate target area based on each axial offset data.

[0102] Specifically, the urethral axis of the prostatic urethra is obtained by fitting the third image data. The frame interval is set according to the detection accuracy requirements. Starting from the bladder neck, the three-dimensional reconstruction model is axially segmented along the transverse plane based on the set frame interval to obtain a continuous sequence of transverse images. The urethral closure contour is extracted from each transverse image in the sequence of transverse images.

[0103] Axial offset data represents the trend information of morphological offset of the prostatic urethra along the urethral axis. For example, axial offset data includes, but is not limited to, the trend information corresponding to the area change rate, centroid offset, or surface elevation change value. Using statistical methods or polygon integration, the contour areas corresponding to two adjacent urethral closure contours are calculated, and the area change rate is determined based on the two contour areas. The geometric centroids corresponding to two adjacent urethral closure contours are extracted, and the Euclidean distance between the two geometric centroids, i.e., the centroid offset, is calculated. For each cross-sectional image, the coordinates of the intersection point of the urethral axis in the cross-sectional image are obtained, and the distance from each boundary point of the urethral closure contour in the cross-sectional image to the coordinates of the intersection point is calculated, i.e., the elevation value. The elevation difference is obtained by subtracting the corresponding two elevation values ​​at each boundary point, and the maximum elevation difference corresponding to each boundary point is taken as the surface elevation change value.

[0104] If the rate of change of area exceeds the corresponding threshold, the prostatic urethra is displaced. If the centroidal offset exceeds the corresponding threshold, the prostatic urethra is contracted or loses radial constraint. If the surface elevation change value exceeds the corresponding threshold, the prostatic urethra has discontinuous edges. If the rate of change of area, centroidal offset, or surface elevation change value all exceed their respective thresholds, the prostatic urethra exhibits a "step" shape or discontinuous edges.

[0105] For example, if a segment of the prostatic urethra shows a sudden decrease in area, a smaller centroid offset, and an increased surface elevation abrupt change, then the segment may have incomplete tissue removal, with residual abnormal tissue causing abrupt morphological changes. If a segment shows a gradual decrease in area and irregular changes in centroid offset, then the segment may have local collapse, leading to deformation. If a segment shows a sudden increase in area, an extremely large surface elevation abrupt change, and irregular changes in centroid offset, then the imaging probe may be interfering, causing voids to appear in the imaging of that segment.

[0106] For example, the axial offset is calculated by aggregation for each axial offset data, and the unit complement of the statistical values ​​corresponding to multiple axial offsets is used as the urethral coherence feature. The statistical value can be the maximum value, average value or median value, etc., but is not limited to the given example.

[0107] In another optional embodiment, the morphological feature layer is an elimination feature layer, and the third image data is a three-dimensional reconstruction model of the prostate target area. The step of extracting features from the input third image data through each morphological feature layer to obtain the structural morphological features of the prostate urethra includes: obtaining the first intraluminal volume of the prostate urethra within the prostate target area through the elimination feature layer; determining the second intraluminal volume based on the third image data; and determining the tissue elimination features corresponding to the prostate target area based on the first intraluminal volume and the second intraluminal volume.

[0108] In this embodiment, the first intracavitary volume is determined based on historical image data corresponding to the prostate target area, which was acquired before the intervention operation was performed on the prostate target area. Specifically, the historical image data is obtained by performing three-dimensional reconstruction processing on second image data, which was obtained using an image acquisition device before the intervention operation was performed on the prostate target area. The image acquisition device used before the intervention operation and the image acquisition device used after the intervention operation can be the same or different.

[0109] In an optional embodiment, the method further includes: sequentially performing coordinate system registration and non-rigid correction on the historical image data and the third image data. Coordinate system registration is used to eliminate overall model offset caused by changes in body position, and non-rigid correction is used to eliminate potential elastic deformation of the tissue.

[0110] In another optional embodiment, the method further includes: acquiring the operation area corresponding to the intervention operation, and performing data cropping on historical image data and third image data according to the operation area. Accordingly, the first intracavitary volume and the second intracavitary volume represent the intracavitary volume of the prostatic urethra within the operation area.

[0111] The advantage of this setting is that it avoids interference from normal volume fluctuations in non-operational areas and solves the problem of increased numerical errors caused by an excessively large evaluation range.

[0112] For example, a triangular mesh volume algorithm or a voxel volume estimation algorithm can be used to determine the first cavity volume and the second cavity volume. The tissue elimination feature can be the cavity volume difference or the cavity volume change rate, where the cavity volume difference represents the amount of abnormal tissue eliminated in the prostatic urethra, but is not limited to the example given above.

[0113] S240. Determine the morphological feature data based on the morphological features of each structure.

[0114] For example, when there are multiple structural morphological features, these features can be classified and organized according to each sub-evaluation model in the structural evaluation model. The sub-evaluation model can be a two-dimensional evaluation model, a three-dimensional evaluation model, a first unobstructed model, a second unobstructed model, or a third unobstructed model.

[0115] S250. The structural evaluation network is used to perform structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target area.

[0116] Based on the above embodiments, optionally, the structural evaluation result further includes a third evaluation result corresponding to each sub-evaluation model, and the structural evaluation network further includes a fusion layer corresponding to each sub-evaluation model. The step of evaluating the prostate target area by means of the structural evaluation network based on the morphological feature data to determine the structural evaluation data further includes: for each sub-evaluation model, performing a fusion evaluation by means of the fusion layer based on the second evaluation result output by at least one structural evaluation layer to obtain the third evaluation result corresponding to the sub-evaluation model.

[0117] Specifically, the third evaluation result includes at least one of the third evaluation score, third evaluation grade, and third evaluation conclusion. For example, the fusion layer can be a logic decision tree, a weighted fusion layer, a machine learning model, or a deep learning network, but is not limited to the examples given above.

[0118] In an optional embodiment, the method further includes: when the structural evaluation model includes a two-dimensional evaluation model and a three-dimensional evaluation model, performing cross-validation on the third evaluation results corresponding to the two-dimensional evaluation model and the three-dimensional evaluation model respectively, and adding the obtained validation results to the structural evaluation results. The third evaluation result is a third evaluation score or a third evaluation level, and the validation results represent the evaluation difference information corresponding to the two-dimensional evaluation model and the three-dimensional evaluation model.

[0119] In existing technologies, quantification operations based on image data often employ holistic measurement techniques without considering the specific structural morphology requirements of the prostate target area. Consequently, the acquired morphological features cannot accurately reflect the true structural morphology of the prostate target area, resulting in technical defects such as information redundancy, neglect of key information, or amplification of measurement errors.

[0120] The technical solution of this embodiment determines the third image data corresponding to each morphological feature layer based on the first image data through the image input layer. Then, each morphological feature layer performs feature extraction based on the input third image data to obtain the structural morphological features of the prostatic urethra. Based on at least one structural morphological feature, morphological feature data is determined. This solves the problem of inaccurate measurement in the overall measurement technology, accurately locks the area range that matches the morphological features, improves the ability of morphological feature data to represent the structural morphology of the prostate, and further improves the accuracy of prostate structure assessment.

[0121] Figure 3This is a flowchart of another prostate image-based assessment method provided in one embodiment of the present disclosure. This embodiment further refines the "performing structural assessment based on the morphological feature data using the structural assessment network to determine the structural assessment data of the prostate target area" in the above embodiment. In this embodiment, the step of performing structural assessment based on the morphological feature data using the structural assessment network to determine the structural assessment data of the prostate target area includes: acquiring a reliable assessment model and assessment object data that match the structural assessment model, wherein the reliable assessment model includes a reliable feature network and a reliable assessment network; extracting features from the assessment object data using the reliable feature network to obtain reliable feature data corresponding to the morphological feature data; filtering the morphological feature data to obtain an invalid feature set based on the reliable assessment network and the reliable feature data; updating the network parameters of the structural assessment network based on the invalid feature set if the invalid feature set is not empty; and performing structural assessment based on the morphological feature data using the updated structural assessment network to determine the structural assessment data of the prostate target area. Figure 3 As shown, the method includes:

[0122] S310. Acquire first image data of the prostate target area that matches the structural evaluation model.

[0123] S310 in this embodiment is the same as that in the above embodiment. Figure 1 The S110 shown is the same as or similar to that in the above embodiments. Figure 2 The S210 shown is the same or similar, and will not be described again in this embodiment.

[0124] Based on the above embodiments, the method may optionally further include: in a cystoscopy imaging scenario, before image acquisition is initiated, adjusting the pump flow rate of the pumping system according to a preset pressure range until the real-time infusion pressure meets the preset pressure range, and then shutting down the pumping system; during image acquisition, in response to the detection of image blurring caused by interfering media, starting the pumping system to synchronously perform infusion and aspiration operations with the same flow rate.

[0125] Infusion pressure refers to the intravesical pressure generated when the infusion fluid is injected into the bladder cavity. It directly affects the degree of expansion and structural stability of the prostatic urethra. Excessive infusion pressure can lead to overexpansion of the prostatic urethra, resulting in an overestimation of its structural features. It may also cause reflux of the infusion fluid, generating air bubbles and causing bright spots or blurring in the image. Conversely, insufficient infusion pressure can lead to insufficient expansion of the prostatic urethra, resulting in wall collapse or overlapping folds, causing chaotic image layers and reducing image clarity and effectiveness. The preset pressure range is adapted to the physiological tolerance range of the prostatic urethra; for example, the preset pressure range can be 60–120 cmH2O, and the infusion pressure can be monitored in real time via a pressure monitoring device.

[0126] Pump flow rate refers to the volume of irrigation fluid delivered to the bladder cavity per unit time by the pumping system. It can be used to regulate irrigation pressure, maintain field of view stability, and ensure the cleanliness of the circulating irrigation fluid. Excessive pump flow rate can cause ripples and disturbances in the irrigation fluid, resulting in ghosting and blurring of the image. Insufficient pump flow rate cannot promptly remove interfering media from the field of view, causing them to accumulate on the cystoscope and resulting in blurred images. The preset pump flow rate range can be adapted to the preset pressure range, such as 100-250 ml / min.

[0127] For example, interfering media may be blood, tissue debris, mucosal secretions, air bubbles, instrument wear debris, and perfusion fluid impurities, but are not limited to the examples given above.

[0128] Based on the above embodiments, optionally, during the image acquisition process, the method further includes: acquiring an environmental parameter sequence in real time according to the parameter acquisition frequency using a parameter sensor, and associating and storing the original image frames acquired by the cystoscope with the environmental parameter sequence, wherein the parameter acquisition frequency is the same as the imaging frame rate of the cystoscope.

[0129] The environmental parameter sequence includes perfusion pressure, pump flow rate, and spatial distance, where spatial distance represents the minimum distance between the cystoscope and the urethral lumen in the cross-sectional plane.

[0130] In an optional embodiment, the method further includes: pausing the image acquisition of the cystoscope in response to the spatial distance not meeting the preset distance range, adjusting the acquisition position of the cystoscope according to the preset distance range, and / or outputting a warning message to indicate that the spatial distance deviates from the preset distance range.

[0131] Specifically, the acquisition position of the cystoscope is adjusted, and in response to the spatial distance meeting a preset distance range, the cystoscope is controlled to continue image acquisition. For example, the output of warning information may include, but is not limited to, text, audio prompts, and indicator lights, and in response to receiving an acquisition restart command, the cystoscope is controlled to continue image acquisition.

[0132] The advantage of this setting is that it maintains the consistency of scale in cystoscopic image acquisition, avoids image scaling distortion due to distance deviation, and thus further ensures the accuracy of prostate structure assessment.

[0133] In an optional embodiment, the method further includes: measuring the distance between the cystoscope and the urethral cavity at the current moment using at least two distance measurement techniques; and aggregating or calibrating the measured distances to obtain the spatial distance at the current moment.

[0134] For example, distance measurement technology can be distance measurement components, visual analysis technology, or calibration measurement technology, but is not limited to the examples given.

[0135] The distance measurement component is integrated into the parameter sensor. Its optical path is coaxially arranged with the optical axis of the cystoscope, ensuring that the component and cystoscope are aligned. The acquisition frequency of the distance measurement component is synchronized with the imaging frame rate of the cystoscope. Exemplary components include, but are not limited to, laser ranging modules, structured light projection modules, and time-of-flight modules. In visual analysis, historical image sequences acquired before the current moment are obtained. The cystoscope's movement trajectory is obtained by processing these sequences using an inter-frame matching algorithm. The marker distances between anatomical markers in the current original image frame are extracted. A linear correlation model between the marker distances and the movement trajectory is established. Using the cystoscope's movement as a constraint, the measured distance is inferred through model iteration. In calibration measurement, the calibration component can be a physical or visual calibration component. Physical calibration components can be reference rods and equidistant rulers, while visual calibration components can be standard circles, grid arrays, scale rods, or other preset calibration patterns. The pixel dimensions of the calibration component in the current original image frame are extracted using an image recognition algorithm, and the measured distance is calculated by combining this with the actual dimensions of the calibration component.

[0136] Based on the above embodiments, the method may optionally further include: during the image acquisition process, inputting the real-time distance at the current moment into the distance calibration model to obtain the output distance correction value, and correcting the real-time distance according to the distance correction value.

[0137] Specifically, the distance calibration model represents the correlation between the calibrated distance and the measured distance. For example, a calibration plate is fixed based on multiple calibrated distances, and a laser emission component is activated to project a calibrated light spot onto the calibration plate. The projection direction of the light spot is coaxial with the optical axis of the cystoscope. The cystoscope acquires an image of the calibration plate at the calibrated distance, and the measured distance obtained through one or more distance measurement techniques is recorded. A distance calibration model is then built based on multiple sets of calibrated distances and measured distances.

[0138] Based on the above embodiments, optionally, the method further includes: acquiring the posture data of the cystoscope in real time through a posture sensor integrated at the end of the cystoscope, determining the optical axis direction of the cystoscope based on the posture data, and storing the posture data, optical axis direction and spatial distance in association to provide a precise spatial reference benchmark for subsequent image information processing.

[0139] S320. The first image data is subjected to feature extraction through the structural morphology network to obtain morphological feature data.

[0140] S320 in this embodiment is the same as that in the above embodiment. Figure 1 The S120 shown is the same as or similar to that in the above embodiments. Figure 2 The S220-S240 shown are the same or similar, and will not be described again in this embodiment.

[0141] Based on the above embodiments, optionally, the structural spatial features in the morphological feature data represent actual measurement information in the physical environment. The structural spatial features are determined by the structural morphology network according to the first scale factor corresponding to the structural spatial features. The structural spatial features are structural morphological features with spatial feature attributes.

[0142] Specifically, structural spatial features represent the quantitative information of the spatial characteristics of the prostate structure. Spatial feature attributes include one-dimensional, two-dimensional, and three-dimensional feature attributes. For example, one-dimensional feature attributes correspond to length, distance, and offset, two-dimensional feature attributes correspond to area, and three-dimensional feature attributes correspond to volume.

[0143] Specifically, the first scale factor is input into the morphological feature layer corresponding to the structural spatial features, so that the morphological feature layer performs scale recovery on the extracted structural morphological features according to the input first scale factor and outputs the structural spatial features. The first scale factor is used to restore the pixel-level parameters to the actual physiological parameters, ensuring the physical authenticity of the structural morphological features.

[0144] In one optional embodiment, the first scale factor is determined based on the scanning parameters of the image acquisition device, including pixel spacing and slice thickness. For example, based on the scanning parameters, the pixel distance between two anatomical feature points in the image data is determined, the actual physiological distance between the two anatomical feature points is obtained, and the ratio of the actual physical distance to the pixel distance is used as the first scale factor.

[0145] In another alternative embodiment, the method further includes: determining environmental parameter data based on a second anatomical region corresponding to the structural spatial feature; and determining a first scale factor corresponding to the structural spatial feature based on the environmental parameter data.

[0146] In this embodiment, the environmental parameter data includes perfusion pressure data, pump flow rate data, and spatial distance data. The spatial distance data represents the minimum distance information between the cystoscope and the urethral wall on the transverse plane. The environmental parameter sequence in the environmental parameter data corresponds one-to-one with the original image frames acquired within the area defined by the second anatomical region.

[0147] For example, scale factor Satisfying the formula: ,in, This indicates the injection pressure data. This indicates pump flow rate data. Represents spatial distance data, This indicates that the correlation model is obtained by training a sample set of environmental parameters.

[0148] Taking the second anatomical region as anatomical region A as an example, the first scale factor corresponding to anatomical region A is expressed as: Correspondingly, the scale recovery coefficient of the structural space feature corresponding to the one-dimensional feature attribute is The scale recovery coefficient of the structural spatial features corresponding to the two-dimensional feature attributes is The scale recovery coefficient of the structural spatial features corresponding to the three-dimensional feature attributes is .

[0149] The advantage of this setup is that it eliminates the dimensional scaling bias caused by factors such as image scanning, 3D reconstruction, spatial registration, and positional differences. It achieves standardization of spatial features under different imaging objects, different image acquisition devices, or different image information processing, ensuring the horizontal consistency of structural spatial features in different prostate assessment tasks and improving the authenticity and comparability of structural spatial features.

[0150] S330. Obtain reliable evaluation model and evaluation object data that match the structure evaluation model.

[0151] In this embodiment, the reliable evaluation model includes a reliable feature network and a reliable evaluation network. The reliable feature network is used to extract feature information that quantifies the reliability of morphological feature data from the evaluation object data, and the reliable evaluation network is used to evaluate the reliability of morphological feature data based on the output of the reliable feature network.

[0152] In an optional embodiment, the evaluation object data includes first image data and / or environmental parameter data corresponding to each structural spatial feature.

[0153] S340. The reliable feature network is used to extract features from the evaluation object data to obtain reliable feature data corresponding to the morphological feature data. Based on the reliable evaluation network and the reliable feature data, the morphological feature data is filtered to obtain an invalid feature set.

[0154] In one alternative embodiment, the reliable feature network includes a first feature layer in the image quality dimension and / or a second feature layer in the scale recovery dimension. Correspondingly, the reliable evaluation network includes a first evaluation layer connected to the first feature layer and / or a second evaluation layer connected to the second feature layer.

[0155] The first feature layer is used to extract image quality features from the first image data. The first evaluation layer quantifies the credibility of morphological feature data by mapping the image quality of the first image data. For example, the image quality dimensions include, but are not limited to, image input quality, image processing quality, and image output quality. The second feature layer is used to extract scale recovery accuracy features from the environmental parameter data of structural spatial features. The second evaluation layer quantifies the credibility of structural spatial features in the morphological feature data by mapping the scale recovery accuracy reflected by the environmental parameter data. For example, the scale recovery dimensions include, but are not limited to, fluctuation dimension, error dimension, and anti-interference dimension.

[0156] In an optional embodiment, the first feature layer includes an image matching module and a feature extraction module. The step of extracting features from the evaluation object data using the reliable feature network to obtain reliable feature data corresponding to the morphological feature data includes: obtaining at least one second anatomical region corresponding to the structural morphology network using the image matching module; obtaining second image data matching each second anatomical region from the first image data; performing feature extraction on each second image data using the feature extraction module to obtain a first feature sequence corresponding one-to-one with the second image data; and determining the reliable feature data corresponding to the morphological feature data based on each of the first feature sequences.

[0157] In this embodiment, the second anatomical region refers to the bladder neck, middle part of the prostate, apex of the prostate, or the entire prostate target area within the prostate target region. Specifically, different structural morphological features may correspond to the same second anatomical region.

[0158] In an optional embodiment, the feature extraction module includes a two-dimensional feature unit and / or a three-dimensional feature unit. The two-dimensional feature unit is used to extract the occlusion area ratio and / or registration residual of the second image data, and the three-dimensional feature unit is used to extract the inter-frame disparity coverage and / or reconstruction residual of the second image data.

[0159] The occlusion area ratio represents the degree to which the visible proportion of the prostate structure decreases due to occlusion by media such as blood, blisters, and mucus in a two-dimensional image sequence. For example, the fourth image data corresponding to the two-dimensional image sequence is obtained. Using image recognition technology, contour recognition is performed on the occluded region formed by the occluding medium and the prostate structure in the second anatomical region of the fourth image data, respectively, to obtain the contours of the occluded region and the structural region. The occlusion area and structural area corresponding to the contours of the occluded region and the structural region are calculated, respectively, and the ratio of the occlusion area to the structural area is taken as the occlusion area ratio.

[0160] The registration residual is an indicator that quantifies the registration accuracy of a two-dimensional image sequence, characterizing the degree of geometric deviation after feature point matching or edge contour alignment in two-dimensional registration processing. Taking feature point matching as an example, matching feature point pairs are extracted from two prostate images, the feature distance between the feature point pairs is calculated, and the statistical value of one or more feature distances is used as the registration residual. The statistical value includes, but is not limited to, the maximum value, minimum value, or average value.

[0161] Inter-frame disparity coverage refers to the degree of disparity redundancy formed by the original image frames from different shooting angles covering the prostate structure in the prostate target area for structural morphology evaluation during 3D reconstruction processing. The higher the inter-frame disparity coverage, the higher the frequency of repeated observation of the prostate structure from multiple perspectives during image acquisition, and the more sufficient the redundancy of disparity information. This can effectively reduce reconstruction errors caused by noise and occlusion in single-frame images, and significantly improve the accuracy and stability of the 3D reconstruction model.

[0162] For example, the fifth image data corresponding to the three-dimensional reconstruction model is obtained, and a set of structural points of the prostate structure in the second anatomical region is defined. The imaging state of each structural point in the set of structural points in each original image frame is traversed. If the structural point is clear and unobstructed, it is considered a valid observation and the imaging state is marked as 1. Otherwise, it is marked as 0. The number of frames covered by the structural point in the fifth image data that are effectively observed is counted. If the number of covered frames is greater than the preset number of redundant frames, it is considered that the structural point has the conditions for reconstruction. Otherwise, it is considered that the structural point does not have the conditions for reconstruction. The number of all structural points that have the conditions for reconstruction is counted, and the ratio of the counted number to the total number of structural points in the set of structural points is used as the inter-frame disparity coverage rate.

[0163] The reconstruction residual is used to quantify the reconstruction quality of the 3D reconstruction model and to quantitatively describe the spatial discreteness of the reconstructed point cloud relative to the reference fitted surface. For example, the reconstruction residual can be defined by at least one of the statistical parameters, namely root mean square error, mean absolute error, maximum residual value, and residual standard deviation. The root mean square error reflects the overall reconstruction deviation of the 3D reconstruction model, the mean absolute error reflects the average reconstruction deviation of the 3D reconstruction model, the maximum residual reflects the local extreme deviation of the 3D reconstruction model, and the residual standard deviation reflects the degree of deviation fluctuation of the 3D reconstruction model.

[0164] In this embodiment, the occlusion area ratio and inter-frame parallax coverage are quantitative features under the image input quality dimension, representing the information integrity of the prostate structure in the original image frame. The registration residual and reconstruction residual are quantitative features under the image processing quality dimension, representing the spatial consistency between the measured features of the prostate structure and the digital representation model.

[0165] Specifically, the reliable feature data includes the first feature data, in which the image quality sequence corresponds one-to-one with the structural morphological features in the morphological feature data.

[0166] Specifically, the first evaluation layer includes a first aggregation model corresponding to each structural morphological feature. In an optional embodiment, the morphological feature data is filtered to obtain an invalid feature set based on the reliable evaluation network and the reliable feature data. This includes: for each structural morphological feature, obtaining an image quality sequence corresponding to the structural morphological feature from the first feature data; aggregating the image quality sequence using the first aggregation model corresponding to the structural morphological feature to obtain a first reliable result corresponding to the structural morphological feature; and filtering the morphological feature data based on each first reliable result to obtain an invalid feature set.

[0167] The image quality sequence includes at least one first feature sequence. For example, the first aggregation model can be a logical decision tree, a weighted fusion model, a machine learning model, or a deep learning network, but is not limited to the given examples.

[0168] Specifically, the first reliable result represents the degree of credibility of the image data in the local dimension, reflecting the local credibility of the first image data. The first reliable result includes at least one of the following: first credibility score, first credibility level, or first credibility conclusion.

[0169] For example, when the first aggregation model includes a two-dimensional hierarchical model, the image quality sequence includes an occlusion area ratio and a registration residual corresponding to the structural morphological features. If the occlusion area ratio is less than a first proportional threshold and the registration residual is less than a first residual threshold, the first reliable result is of high confidence. If the occlusion area ratio is within a first proportional range defined by the first proportional threshold and the second proportional threshold, or the registration residual is within a first residual value range defined by the first residual threshold and the second residual threshold, the first reliable result is of medium confidence. If the occlusion area ratio is greater than the second proportional threshold and the registration residual is greater than the second residual threshold, the first reliable result is of low confidence. Wherein, the first proportional threshold is less than the second proportional threshold, and the first residual threshold is less than the second residual threshold. For example, the first proportional range can be [15%, 40%], and the first residual value range can be [1mm, 2mm], but it is not limited to the given example.

[0170] When the first aggregation model includes a three-dimensional hierarchical model, the image quality sequence includes inter-frame disparity coverage and reconstruction residual corresponding to structural morphological features. If the image quality value of the inter-frame disparity coverage is greater than a third proportional threshold and the reconstruction residual is less than a third residual threshold, the first reliable result is of high confidence. If the image quality value of the inter-frame disparity coverage is within the second proportional range defined by the third and fourth proportional thresholds, or if the reconstruction residual is within the second residual range defined by the third and fourth residual thresholds, the first reliable result is of medium confidence. If the image quality value of the inter-frame disparity coverage is less than a fourth proportional threshold and the reconstruction residual is greater than a fourth residual threshold, the first reliable result is of low confidence. Wherein, the third proportional threshold is greater than the fourth proportional threshold, and the third residual threshold is less than the fourth residual threshold. For example, the second proportional range can be [60%, 85%], and the third residual range can be [1mm, 2mm], but it is not limited to the given example.

[0171] In another optional embodiment, the step of extracting features from the evaluation object data through the reliable feature network to obtain reliable feature data corresponding to the morphological feature data includes: determining a second feature sequence based on each environmental parameter data through the second feature layer, and determining reliable feature data of the morphological feature data based on each second feature sequence.

[0172] In an optional embodiment, the second feature sequence includes scale fluctuation features and / or scale offset features.

[0173] The scale fluctuation characteristic is used to reflect the stability of the first scale factor. Specifically, an instantaneous scale factor is determined for each environmental parameter sequence in the environmental parameter data, resulting in a scale factor sequence composed of each instantaneous scale factor. Fluctuation analysis is then performed on the scale factor sequence to obtain the scale fluctuation characteristic. For example, the fluctuation analysis method can be standard deviation, coefficient of variation, maximum-minimum range, or analysis of variance, etc., but is not limited to the given examples.

[0174] The scale offset feature is used to reflect the systematic distortion of the first scale factor. Specifically, the environmental parameter data is subjected to fluctuation analysis to obtain environmental fluctuation data. If the environmental fluctuation data does not meet the allowable fluctuation conditions, the second scale factor corresponding to the structural spatial feature is obtained. The error feature value between the first scale factor and the second scale factor is used as the scale offset feature, wherein the error feature value is the error value or the error rate.

[0175] For example, environmental fluctuation data includes the number of times environmental parameters exceed the preset parameter range, the standard deviation or gradient of environmental parameters, and environmental parameters include injection pressure, pump flow rate and spatial distance.

[0176] In a specific example, obtaining the second scale factor corresponding to the structural spatial features includes: obtaining third image data matching the structural spatial features; extracting pixel size data of the calibration component from the third image data; and determining the second scale factor based on the pixel size data and the actual physical size of the calibration component; or, extracting feature point pairs with known physical distances from the third image data matching the structural spatial features; determining pixel distance data corresponding to each feature point pair; and determining inter-frame transformation data based on the third image data matching the structural spatial features; obtaining cystoscope pose data matching the structural spatial features; constructing an objective function based on the physical distance data, pixel distance data, inter-frame transformation data, and cystoscope pose data; wherein the optimization variables of the objective function include the scale factor and the cystoscope pose; and iteratively optimizing the objective function to obtain the second scale factor.

[0177] Specifically, reliable feature data includes second feature data, and the second feature sequence in the second feature data corresponds one-to-one with the structural spatial features in the morphological feature data.

[0178] Specifically, the second evaluation layer includes a second aggregation model corresponding to each structural spatial feature. In an optional embodiment, the morphological feature data is filtered to obtain an invalid feature set based on the reliable evaluation network and the reliable feature data. This includes: for each structural spatial feature, obtaining a second feature sequence corresponding to the structural spatial feature from the reliable feature data; aggregating the second feature sequence using the second aggregation model corresponding to the structural spatial feature to obtain a second reliable result corresponding to the structural spatial feature; and filtering the morphological feature data based on each second reliable result to obtain an invalid feature set.

[0179] Specifically, the second reliable result represents the credibility of the structural spatial features in the physical environment. The second reliable result includes at least one of a second credibility score, a second credibility level, or a second credibility conclusion. For example, when the second aggregation model is a hierarchical model, if the scale fluctuation feature is greater than the fluctuation threshold or the scale offset feature is greater than the offset threshold, the second reliable result is of low credibility; if the scale fluctuation feature is less than the fluctuation threshold and the scale offset feature is less than the offset threshold, the second reliable result is of high credibility.

[0180] Specifically, the invalid feature set includes structural morphological features whose first reliable result and / or second reliable result do not meet the credible evaluation condition. For example, if the invalid feature set is empty, it means that each structural morphological feature in the morphological feature data meets its corresponding credible evaluation condition. If the invalid feature set is not empty, it means that at least one structural morphological feature in the morphological feature data does not meet its corresponding credible evaluation condition. The credible evaluation condition can be a range of valid results or a threshold for valid results.

[0181] S350. If the invalid feature set is not empty, update the network parameters of the structure evaluation network according to the invalid feature set.

[0182] In an optional embodiment, updating the network parameters of the structure evaluation network based on the invalid feature set includes: obtaining reliable result data corresponding to the invalid feature set, and updating the network parameters of the structure evaluation network based on the reliable result data. The reliable result data includes a first reliable result and / or a second reliable result, and the reliable result includes a confidence score and / or a confidence level.

[0183] For example, updating the network parameters of a structural evaluation network can be done by reducing the weight parameters corresponding to invalid structural morphological features in the structural evaluation network, or by deleting the feature parameters corresponding to invalid structural morphological features in the structural evaluation network, but is not limited to the given example.

[0184] S360. The structure is evaluated based on the morphological feature data using the updated structure evaluation network to determine the structure evaluation data of the prostate target area.

[0185] S360 in this embodiment is the same as in the above embodiment. Figure 1 The S130 shown is the same as or similar to that in the above embodiments. Figure 2 The S250 shown is the same or similar, and will not be described again in this embodiment.

[0186] Based on the above embodiments, optionally, the structural evaluation data also includes reliable evaluation data corresponding to the structural evaluation results, such as image reliability results corresponding to the first evaluation result, first reliable results corresponding to each second evaluation result, second reliable results of the second evaluation results corresponding to the structural spatial features, and third reliable results corresponding to the third evaluation result, etc. The image reliability results are obtained by aggregating and analyzing all the first reliable results, reflecting the overall credibility information of the first image data. The image reliability results include at least one of image credibility score, image credibility level, and image credibility conclusion.

[0187] Poor image quality can lead to distortion of structural morphological features and the generation of meaningless redundant features. Poor scale recovery accuracy can superimpose scale scaling distortion on structural spatial features, causing the structural spatial features to deviate significantly from the actual physiological scale, ultimately resulting in insufficient accuracy in prostate structure assessment.

[0188] The technical solution of this embodiment builds a reliable evaluation model from the dimensions of image quality and scale recovery. Based on the reliable evaluation model, it performs reliability evaluation on morphological feature data to achieve feature screening. Based on the set of invalid features selected, it updates the network parameters of the structural evaluation network. The updated structural evaluation network performs structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target area. This solves the problem of prostate evaluation task failure caused by inaccurate structural morphological features, reduces the decision pollution caused by invalid features in prostate structural evaluation, makes the structural evaluation results more realistic and reliable, breaks through the performance bottleneck of prostate evaluation task in complex image acquisition environment, improves the robustness of structural evaluation model, and provides effective assistance for promoting the standardized iteration of prostate intervention technology.

[0189] It should be noted that the above embodiments of this disclosure are mainly illustrated by taking the assessment needs of the prostatic urethra covered by the structural assessment model as an example. It is understood that the structural assessment model can also be used to assess the needs of other anatomical parts of the prostate, such as the acinar-duct system or the combined structure of the verumontanum and ejaculatory duct, to provide data support for the analysis and determination of the secretory or reproductive functions of the prostate. Of course, the structural assessment model can also cover multiple assessment needs at the same time to achieve multi-dimensional collaborative assessment.

[0190] Example 1: Comprehensive Evaluation Report

[0191] In the application scenario of real-time operational assessment of ablation procedures, the structural assessment model covers the assessment needs of the prostatic urethra. The comprehensive assessment report, compiled from the structural assessment data, includes textual content such as a comprehensive assessment summary, a feature assessment summary, a quality control summary, and interpretations and recommendations. The comprehensive assessment summary is a structured text generated based on the first assessment result; the feature assessment summary is a structured text generated based on one or more second assessment results, which may include normalized scores of structural morphological features, descriptive text corresponding to the second assessment level, etc.; the quality control summary represents the prompts for the second assessment results, such as anomaly warnings, risk warnings, and reliability warnings; and the interpretations and recommendations are textual interpretations of the assessment reliability generated based on reliable assessment data, as well as suggestive text to guide operational optimization, including image acquisition operation optimization and / or ablation operation optimization.

[0192] For example, the comprehensive assessment summary: the structural score of the prostatic urethra, P=0.675, is at a moderate level, indicating good immediate improvement from the ablation procedure, and the reliability of the structural score P is moderate; the feature assessment summary: the bladder neck opening angle C1=78°, the initial opening is adequate, and the bladder neck opening area C2=30mm². 2 The prostate apex showed no obvious deformation, with an opening angle C3 of 60°. The distal opening narrowed slightly, and the opening area C4 was 39 mm². 2 The distal opening was not adhered, the urethral symmetry feature C5=1.5°, the urethral lumen was symmetrical and uniform from left to right, the urethral continuity feature C6=0.2mm, the lumen was continuous and uninterrupted, the tissue removal volume ΔV=8.4ml, and the tissue resection was sufficient.

[0193] Quality Control Summary: Bladder neck opening angle C1: No abnormalities, no potential risks, [Confidence Hint] Bladder neck incomplete, feature assessment requires manual review; Bladder neck opening area C2: No abnormalities, no potential risks, [Confidence Hint] Bladder neck incomplete, and dimensional recovery unstable, feature assessment requires manual review; [Abnormality Hint] Prostate apex opening angle C3: Slightly deviates from the normal range, [Risk Hint] Risk of stenosis exists, feature assessment is reliable; Prostate apex opening area C4: No abnormalities, no potential risks, feature assessment is reliable; Urethral symmetry feature C5: No abnormalities in the middle of the prostate, no potential risks, feature assessment is reliable; Urethral continuity feature C6: No distortions, no potential risks, [Confidence Hint] Bladder neck incomplete, feature assessment may be distorted; Tissue elimination amount ΔV: No residue, no potential risks, [Confidence Hint] Bladder neck dimensional recovery unstable, feature assessment may be distorted.

[0194] Interpretation and suggestions: 1) If C1 and C2 features are normal but affected by incomplete radial extension structures and unstable scale recovery, manual review is required to prevent distortion; 2) Optimize the imaging parameters of the image acquisition equipment, such as increasing the number of acquisition frames at the bladder neck or improving the imaging resolution to improve the integrity of the radial extension structures; 3) Reduce the ablation depth and ablation energy at the apex of the prostate to reduce tissue contracture caused by thermal damage.

[0195] Example 2: Multi-dimensional assessment report

[0196] Example 2, based on Example 1, supplements the multidimensional assessment summary of the prostatic urethra in terms of radial patency, axial patency, and spatial patency. For example, the multidimensional assessment summary shows: partial damage to radially extending structures, no abnormalities in the bladder neck, and a risk of stenosis at the apex of the prostate; the prostatic urethra is regular, with a continuous and undistorted morphology; and tissue ablation is complete with no residue.

[0197] The following are embodiments of the prostate image-based evaluation device provided in this disclosure. This device and the prostate image-based evaluation method described above belong to the same inventive concept. For details not described in detail in the embodiments of the prostate image-based evaluation device, please refer to the content of the prostate image-based evaluation method in the above embodiments.

[0198] Figure 4 This is a schematic diagram of the structure of a prostate image-based evaluation device provided in one embodiment of this disclosure. Figure 4 As shown, the device includes: a first image data acquisition module 410, a morphological feature data determination module 420, and a structural evaluation data determination module 430.

[0199] The first image data acquisition module 410 is used to acquire first image data of the prostate target area that matches the structural evaluation model. The structural evaluation model includes a structural morphology network and a structural evaluation network.

[0200] The morphological feature data determination module 420 is used to extract features from the first image data through the structural morphology network to obtain morphological feature data, wherein the morphological feature data represents the quantitative information of the structural morphology of the prostate target area.

[0201] The structural assessment data determination module 430 is used to determine the structural assessment data of the prostate target area by performing structural assessment based on the morphological feature data through the structural assessment network.

[0202] The technical solution of this embodiment constructs a structural assessment model by planning and quantifying features from the structural morphology dimension of the prostate. Through the structural morphology network in the structural assessment model, features are extracted from the first image data of the prostate target area to obtain morphological feature data. The structural assessment network in the structural assessment model evaluates the morphological feature data to determine the structural assessment data of the prostate target area. By utilizing the standardized quantitative representation of the prostate's structural morphology in imaging, the objectivity and repeatability of the prostate assessment task are ensured, solving the problem of insufficient accuracy of assessment models that rely on manual methods. This provides accurate data support for the analysis and judgment of prostate function, and further contributes to promoting the standardized iteration of prostate intervention technology.

[0203] In an optional embodiment, the structural evaluation model includes a first evaluation model and / or a second evaluation model, wherein the first evaluation model is used to evaluate the structural properties of the prostate target area in two-dimensional imaging mode and / or three-dimensional imaging mode, and the second evaluation model is used to evaluate the structural properties of the prostate urethra within the prostate target area in one or more patency characteristic dimensions.

[0204] In one optional embodiment, the second evaluation model includes at least one of a first accessibility model, a second accessibility model, and a third accessibility model;

[0205] When the structural evaluation model includes a first unobstructed model, the structural morphology network includes a geometric feature layer corresponding to the first anatomical region; when the structural evaluation model includes a second unobstructed model, the structural morphology network includes a symmetry feature layer and / or a coherence feature layer; when the structural evaluation model includes a third unobstructed model, the structural morphology network includes an elimination feature layer.

[0206] When the structural evaluation model includes a first evaluation model, the structural morphology network includes a symmetry feature layer and / or a geometric feature layer corresponding to the first anatomical region; when the first evaluation model includes a three-dimensional evaluation model, the structural morphology network further includes a coherent feature layer and / or an elimination feature layer.

[0207] The first anatomical region includes the bladder neck and / or the apex of the prostate in the prostate target area, and the urethral geometric features output by the geometric feature layer represent quantitative information on the opening morphology of the prostate urethra in the first anatomical region.

[0208] In an optional embodiment, the structural evaluation data determination module 430 includes:

[0209] A reliable evaluation model acquisition unit is used to acquire reliable evaluation model and evaluation object data that match the structure evaluation model. The reliable evaluation model includes a reliable feature network and a reliable evaluation network.

[0210] An invalid feature set filtering unit is used to extract features from the evaluation object data through the reliable feature network to obtain reliable feature data corresponding to the morphological feature data, and to filter the morphological feature data to obtain an invalid feature set based on the reliable evaluation network and the reliable feature data.

[0211] The structure evaluation network update unit is used to update the network parameters of the structure evaluation network according to the invalid feature set when the invalid feature set is a non-empty set.

[0212] The structural assessment data determination unit is used to perform structural assessment based on the morphological feature data through an updated structural assessment network to determine the structural assessment data of the prostate target region.

[0213] In an optional embodiment, the evaluation object data includes first image data, and the reliable feature network includes a first feature layer, which includes an image matching module and a feature extraction module.

[0214] The invalid feature set filtering unit includes:

[0215] The second image data acquisition subunit is used to acquire at least one second anatomical region corresponding to the structural morphology network through the image matching module, and to acquire second image data matching each second anatomical region from the first image data;

[0216] The feature extraction module extracts features from each second image data to obtain a first feature sequence that corresponds one-to-one with the second image data.

[0217] Based on each of the first feature sequences, determine the reliable feature data corresponding to the morphological feature data;

[0218] The second anatomical region is the bladder neck, middle part of the prostate, apex of the prostate, or the entire prostate target area within the prostate target area.

[0219] In an optional embodiment, the structural morphology network further includes an image input layer and a morphological feature data determination module 420, comprising:

[0220] The structural morphology feature determination unit is used to determine the third image data corresponding to each morphology feature layer based on the first image data through the image input layer.

[0221] The structural morphology feature determination unit is used to extract features from the input third image data through each morphology feature layer to obtain the structural morphology features of the prostatic urethra.

[0222] The morphological feature data determination unit is used to determine morphological feature data based on the morphological features of each of the structures.

[0223] In an optional embodiment, the morphological feature layer is a geometric feature layer, which includes an angle extraction module and / or an area extraction module, and the third image data is image data corresponding to the bladder neck or the tip of the prostate.

[0224] Structural morphology characteristic determination unit, including:

[0225] A urethral geometric feature determination subunit is used to determine at least one pair of first urethral contours corresponding to the prostatic urethra based on the third image data using the angle extraction module, and to determine the opening angle corresponding to the first anatomical region based on each pair of first urethral contours, wherein the two first urethral contours in the first urethral contour pair are distributed on opposite sides along the urethral axis; and / or,

[0226] The area extraction module determines at least one urethral closure contour corresponding to the prostatic urethra based on the third image data, and determines the opening area corresponding to the first anatomical region based on each urethral closure contour.

[0227] In an optional embodiment, the morphological feature layer is a symmetry feature layer, and the third image data includes image data corresponding to at least one second anatomical region, wherein the second anatomical region is the bladder neck, middle part of the prostate, apex of the prostate, or the entire prostate target area in the prostate target area.

[0228] Structural morphology characteristic determination unit, including:

[0229] The urethral symmetry feature determination subunit is used to obtain second image data that matches the second anatomical region in the third image data for each second anatomical region through the symmetry feature layer;

[0230] Based on the second image data, the longitudinal central axis of the prostatic urethra within the second anatomical region is determined;

[0231] Based on the longitudinal central axis and the second image data, a mirror comparison operation is performed to obtain the urethral symmetry features corresponding to the second anatomical region.

[0232] In an optional embodiment, the morphological feature layer is a coherent feature layer, the third image data is a three-dimensional reconstruction model of the prostate target area, and the structural morphological feature determination unit includes:

[0233] The urethral continuity feature determination subunit is used to determine a continuous sequence of transverse images based on the third image data through the continuity feature layer, and to determine axial offset data based on every two adjacent transverse images in the transverse image sequence.

[0234] Based on the axial offset data, the urethral continuity features corresponding to the prostate target area are determined.

[0235] In an optional embodiment, the morphological feature layer is an elimination feature layer, the third image data is a three-dimensional reconstruction model of the prostate target area, and the structural morphological feature determination unit includes:

[0236] The tissue elimination feature determination subunit is used to obtain the first intraluminal volume of the prostatic urethra within the prostate target area through the elimination feature layer, determine the second intraluminal volume based on the third image data, and determine the tissue elimination feature corresponding to the prostate target area based on the first intraluminal volume and the second intraluminal volume.

[0237] The first cavity volume is determined based on historical image data of the prostate target area, which was acquired before the intervention operation was performed on the prostate target area.

[0238] In an optional embodiment, the structural spatial features in the morphological feature data represent actual measurement information in the physical environment. The structural spatial features are determined by the structural morphology network based on the first scale factor corresponding to the structural spatial features. The structural spatial features are structural morphological features with spatial feature attributes.

[0239] In an optional embodiment, the device further includes:

[0240] The first scale factor determination module is used to determine environmental parameter data based on the second anatomical region corresponding to the structural spatial features. The environmental parameter data includes perfusion pressure data, pump flow rate data, and spatial distance data.

[0241] Based on the environmental parameter data, determine the first scale factor corresponding to the structural spatial characteristics;

[0242] The spatial distance data represents the minimum distance between the cystoscope and the urethral wall in the cross-sectional plane.

[0243] In an optional embodiment, the device further includes:

[0244] The image acquisition control module is used to adjust the pump flow rate of the pumping system according to a preset pressure range before image acquisition is started in a cystoscopy imaging scenario, until the real-time infusion pressure meets the preset pressure range, and then shut down the pumping system.

[0245] During image acquisition, an environmental parameter sequence is collected in real time using a parameter sensor according to the parameter acquisition frequency. The original image frames acquired by the cystoscope are associated with and stored with the environmental parameter sequence. The parameter acquisition frequency is the same as the imaging frame rate of the cystoscope.

[0246] In response to the detection of image blurring caused by interfering media, the pumping system is activated to simultaneously perform injection and suction operations with the same flow rate.

[0247] In an optional embodiment, the evaluation object data includes environmental parameter data corresponding to each structural spatial feature, the reliable feature network includes a second feature layer, and the invalid feature set filtering unit includes:

[0248] The reliable feature data determination subunit determines the reliable feature data of the morphological feature data based on the second feature layer for each environmental parameter data, and determines the second feature sequence based on each second feature sequence.

[0249] The second feature sequence includes scale fluctuation features and / or scale shift features.

[0250] The prostate image-based evaluation device provided in this disclosure can execute the prostate image-based evaluation method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the method.

[0251] Figure 5This is a schematic diagram of an electronic device provided according to one embodiment of the present disclosure. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0252] like Figure 5 As shown, the electronic device 10 includes processors 11 and memory, such as read-only memory (ROM) 12 and random access memory (RAM) 13, communicatively connected to each processor 11. The memory stores computer programs executable by each processor 11. The processor 11 can perform various appropriate actions and processes based on the computer programs stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processors 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0253] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information or data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0254] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the prostate image-based evaluation method provided in the above embodiments.

[0255] In some embodiments, the prostate image-based evaluation method provided in the above embodiments can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the prostate image-based evaluation method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the prostate image-based evaluation method by any other suitable means (e.g., by means of firmware).

[0256] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of embodiments of this disclosure.

[0257] Various embodiments of the systems and techniques described above can be implemented in the following systems or combinations thereof: digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard parts (ASSPs), system-on-chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including each of the described programmable processors, which may be dedicated or general-purpose programmable processors, capable of receiving data and instructions from a storage system, each of the described input devices, and each of the described output devices, and transmitting data and instructions to the storage system, the respective input devices, and the respective output devices.

[0258] Computer programs for implementing the prostate image-based evaluation method of this disclosure can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0259] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable storage medium. Examples of machine-readable storage media include electrical connections based on the aforementioned lines, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0260] To provide interaction with a user, the systems and techniques described herein can be implemented on a terminal device having: a display device for displaying information to the user (e.g., a cathode-ray tube (CRT) or liquid crystal display (LCD) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the terminal device. Other types of devices can also provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0261] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0262] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product in the cloud computing service model to address the shortcomings of traditional physical hosts and Virtual Private Server (VPS) services, such as high management difficulty and weak business scalability.

[0263] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this disclosure can be achieved, and this is not limited herein.

[0264] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An assessment method based on prostate images, characterized in that, include: First image data of the prostate target region matching the structural evaluation model, which includes a structural morphology network and a structural evaluation network, are acquired. The first image data is subjected to feature extraction through the morphological network to obtain morphological feature data, which represents the quantitative information of the structural morphology of the prostate target area. The structural evaluation network is used to perform structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target area; The structural evaluation model includes a first evaluation model and / or a second evaluation model. The first evaluation model is used to evaluate the structural properties of the prostate target area in two-dimensional imaging mode and / or three-dimensional imaging mode, and the second evaluation model is used to evaluate the structural properties of the prostate urethra within the prostate target area in one or more patency characteristic dimensions.

2. The method according to claim 1, characterized in that, The second assessment model includes at least one of the first accessibility model, the second accessibility model, and the third accessibility model; When the structural evaluation model includes a first unobstructed model, the structural morphology network includes a geometric feature layer corresponding to the first anatomical region; when the structural evaluation model includes a second unobstructed model, the structural morphology network includes a symmetry feature layer and / or a coherence feature layer; when the structural evaluation model includes a third unobstructed model, the structural morphology network includes an elimination feature layer. When the structural evaluation model includes a first evaluation model, the structural morphology network includes a symmetry feature layer and / or a geometric feature layer corresponding to the first anatomical region; when the first evaluation model includes a three-dimensional evaluation model, the structural morphology network further includes a coherent feature layer and / or an elimination feature layer. The first anatomical region includes the bladder neck and / or the apex of the prostate in the prostate target area. The urethral geometric features output by the geometric feature layer represent the quantitative information of the opening morphology of the prostate urethra in the first anatomical region. The symmetry feature layer and the coherence feature layer are used to extract the quantitative features of the axial morphology of the prostate urethra. The urethral symmetry features output by the symmetry feature layer represent the relative symmetry of the two urethral halves obtained by axially dividing the prostate urethra. The urethral coherence features output by the coherence feature layer represent the morphological continuity of the prostate urethra along the urethral axis.

3. The method according to claim 1, characterized in that, The step of determining the structural assessment data of the prostate target region by performing structural assessment based on the morphological feature data through the structural assessment network includes: Obtain reliable evaluation model and evaluation object data that match the structural evaluation model, wherein the reliable evaluation model includes a reliable feature network and a reliable evaluation network; The reliable feature network is used to extract features from the evaluation object data to obtain reliable feature data corresponding to the morphological feature data. Based on the reliable evaluation network and the reliable feature data, the morphological feature data is filtered to obtain an invalid feature set. If the invalid feature set is not empty, the network parameters of the structure evaluation network are updated according to the invalid feature set. The updated structural evaluation network is used to perform structural evaluation based on the morphological feature data to determine the structural evaluation data of the prostate target area.

4. The method according to claim 3, characterized in that, The evaluation object data includes first image data, and the reliable feature network includes a first feature layer, which includes an image matching module and a feature extraction module. The step of extracting features from the evaluation object data using the reliable feature network to obtain reliable feature data corresponding to the morphological feature data includes: The image matching module obtains at least one second anatomical region corresponding to the structural morphology network, and obtains second image data matching each second anatomical region from the first image data; The feature extraction module extracts features from each second image data to obtain a first feature sequence that corresponds one-to-one with the second image data. Based on each of the first feature sequences, determine the reliable feature data corresponding to the morphological feature data; The second anatomical region is the bladder neck, middle part of the prostate, apex of the prostate, or the entire prostate target area within the prostate target area.

5. The method according to claim 2, characterized in that, The morphological network further includes an image input layer, wherein feature extraction of the first image data through the morphological network to obtain morphological feature data includes: The image input layer determines the third image data corresponding to each morphological feature layer based on the first image data. By extracting features from the input third image data at each morphological feature layer, the structural morphological features of the prostatic urethra are obtained. Based on the structural morphological characteristics described above, determine the morphological feature data; The morphological feature layer can be the geometric feature layer, the symmetry feature layer, the continuity feature layer, or the elimination feature layer.

6. The method according to claim 5, characterized in that, The morphological feature layer is a geometric feature layer, which includes an angle extraction module and / or an area extraction module. The third image data is image data corresponding to the bladder neck or the tip of the prostate. The structural morphological features of the prostatic urethra are obtained by extracting features from the input third image data at each morphological feature layer, including: The angle extraction module determines at least one pair of first urethral contours corresponding to the prostatic urethra based on the third image data, and determines the opening angle corresponding to the first anatomical region based on each pair of first urethral contours, wherein the two first urethral contours in the first urethral contour pair are distributed on opposite sides along the urethral axis; and / or, The area extraction module determines at least one urethral closure contour corresponding to the prostatic urethra based on the third image data, and determines the opening area corresponding to the first anatomical region based on each urethral closure contour.

7. The method according to claim 5, characterized in that, The morphological feature layer is a symmetrical feature layer, and the third image data includes image data corresponding to at least one second anatomical region, wherein the second anatomical region is the bladder neck, middle part of the prostate, apex of the prostate, or the entire prostate target area in the prostate target area. The structural morphological features of the prostatic urethra are obtained by extracting features from the input third image data at each morphological feature layer, including: For each second anatomical region, the symmetric feature layer is used to obtain second image data in the third image data that matches the second anatomical region; Based on the second image data, the longitudinal central axis of the prostatic urethra within the second anatomical region is determined; Based on the longitudinal central axis and the second image data, a mirror comparison operation is performed to obtain the urethral symmetry features corresponding to the second anatomical region.

8. The method according to claim 5, characterized in that, The morphological feature layer is a coherent feature layer, and the third image data is a three-dimensional reconstruction model of the prostate target area. The structural morphological features of the prostate urethra are obtained by extracting features from the input third image data through each morphological feature layer, including: Based on the third image data, the continuous feature layer determines a continuous sequence of cross-sectional images, and the axial offset data is determined based on every two adjacent cross-sectional images in the sequence of cross-sectional images. Based on the axial offset data, the urethral continuity features corresponding to the prostate target area are determined.

9. The method according to claim 5, characterized in that, The morphological feature layer is a feature elimination layer, and the third image data is a three-dimensional reconstruction model of the prostate target area. The structural morphological features of the prostate urethra are obtained by extracting features from the input third image data through each morphological feature layer, including: The first intraluminal volume of the prostatic urethra within the prostatic target area is obtained through the elimination feature layer. The second intraluminal volume is determined based on the third image data. The tissue elimination feature corresponding to the prostatic target area is determined based on the first intraluminal volume and the second intraluminal volume. The first cavity volume is determined based on historical image data of the prostate target area, which was acquired before the intervention operation was performed on the prostate target area.

10. The method according to claim 3, characterized in that, The structural spatial features in the morphological feature data represent actual measurement information in the physical environment. The structural spatial features are determined by the structural morphology network based on the first scale factor corresponding to the structural spatial features. The structural spatial features are structural morphological features with spatial feature attributes.

11. The method according to claim 10, characterized in that, The method further includes: Based on the second anatomical region corresponding to the structural spatial features, environmental parameter data are determined, including perfusion pressure data, pump flow rate data, and spatial distance data. Based on the environmental parameter data, determine the first scale factor corresponding to the structural spatial characteristics; The spatial distance data represents the minimum distance between the cystoscope and the urethral wall in the cross-sectional plane.

12. The method according to claim 11, characterized in that, The method further includes: In a cystoscopy imaging scenario, before image acquisition is initiated, the pump flow rate of the pumping system is adjusted according to a preset pressure range until the real-time infusion pressure meets the preset pressure range, at which point the pumping system is shut down. During image acquisition, environmental parameter sequences are collected in real time using a parameter sensor according to the parameter acquisition frequency. The original image frames acquired by the cystoscope are associated with and stored with the environmental parameter sequences. The parameter acquisition frequency is the same as the imaging frame rate of the cystoscope. In response to the detection of image blurring caused by interfering media, the pumping system is activated to simultaneously perform injection and suction operations with the same flow rate.

13. The method according to claim 11, characterized in that, The evaluation object data includes environmental parameter data corresponding to each structural spatial feature. The reliable feature network includes a second feature layer. The process of extracting features from the evaluation object data using the reliable feature network to obtain reliable feature data corresponding to the morphological feature data includes: Based on each environmental parameter data, the second feature layer determines a second feature sequence, and based on each second feature sequence, determines reliable feature data of the morphological feature data. The second feature sequence includes scale fluctuation features and / or scale shift features.

14. An assessment device based on prostate images, characterized in that, include: The first image data acquisition module is used to acquire first image data of the prostate target area that matches the structural evaluation model, wherein the structural evaluation model includes a structural morphology network and a structural evaluation network. The morphological feature data determination module is used to extract features from the first image data through the structural morphology network to obtain morphological feature data, wherein the morphological feature data represents the quantitative information of the structural morphology of the prostate target area; The structural assessment data determination module is used to perform structural assessment based on the morphological feature data through the structural assessment network to determine the structural assessment data of the prostate target region. The structural evaluation model includes a first evaluation model and / or a second evaluation model. The first evaluation model is used to evaluate the structural properties of the prostate target area in two-dimensional imaging mode and / or three-dimensional imaging mode, and the second evaluation model is used to evaluate the structural properties of the prostate urethra within the prostate target area in one or more patency characteristic dimensions.

15. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, which enables the at least one processor to perform the prostate image-based evaluation method according to any one of claims 1-13.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the prostate image-based evaluation method according to any one of claims 1-13.

17. A computer program product comprising a computer program that, when executed by a processor, implements the prostate image-based evaluation method according to any one of claims 1-13.

Images