Pathology auxiliary diagnosis system based on image analysis

Abstract

The invention relates to the technical field of medical information processing, in particular to a pathological auxiliary diagnosis system based on image analysis, which comprises a data acquisition module, a pathological analysis module and a pathological auxiliary diagnosis module, the method comprises the following steps: acquiring patient blood pathological region image data through a data acquisition module, dividing the acquired image data into a plurality of regions by utilizing a segmentation algorithm so as to analyze pathological and non-pathological regions of different degrees, and analyzing the pathological and non-pathological regions of different degrees according to the regions divided by the data acquisition module; the pathology analysis module carries out multi-angle pathology analysis according to the pathology area images of different degrees and sends the multi-angle pathology analysis to the pathology auxiliary diagnosis module, and the pathology auxiliary diagnosis module carries out pathology assistance on medical diagnosis according to the received medical record analysis result; through a multi-layer convolutional network and a segmentation algorithm, accurate segmentation and analysis of a pathological image area are improved; accurate analysis is carried out according to the characteristics of different pathological areas, and division and diagnosis of the pathological areas are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information processing, and in particular to a pathology auxiliary diagnosis system based on image analysis. Background Art

[0002] Pathological diagnosis, as the "gold standard" for disease diagnosis, has irreplaceable significance in clinical medicine. By analyzing the morphological characteristics of patient tissues or cells, pathological diagnosis can accurately determine the nature, extent, and prognosis of lesions, providing a core basis for formulating treatment plans. However, traditional pathological diagnosis is highly dependent on the experience accumulated by pathologists and manual microscopic examinations, and has problems such as strong subjectivity, low efficiency, and poor repeatability. Especially when dealing with complex or large sample size pathological images, manual analysis is prone to misdiagnosis and missed diagnosis due to visual fatigue, and it is difficult to achieve dynamic tracking and quantitative evaluation of the lesion area.

[0003] Although existing image analysis technologies have improved the efficiency of pathological diagnosis to a certain extent, they still face significant technical bottlenecks: (1) Traditional image segmentation algorithms (such as thresholding and edge detection) have poor adaptability to complex pathological images and are difficult to accurately segment the lesion area; (2) Static analysis models cannot capture the dynamic evolution characteristics of the lesion area, resulting in delayed disease assessment; (3) There is a lack of a multi-dimensional parameter fusion mechanism, and the diagnosis results rely on a single indicator (such as area or concentration), which affects the accuracy of the comprehensive judgment. These problems have seriously restricted the clinical application value of pathology-assisted diagnosis systems. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a pathology auxiliary diagnosis system based on image analysis, comprising:

[0006] The data acquisition module is used to collect images of the patient's blood pathology area, specifically:

[0007] By collecting the patient's pathological image data and using the segmentation algorithm to divide the image data into three regions, the image data is accurately segmented according to the probability of the pixel points in each category, including:

[0008] By constructing a multi-layer convolutional network structure, convolution operations are performed layer by layer, and finally the convolution operation results are converted into the probability of the corresponding category;

[0009] The pathology analysis module is used to perform multi-angle pathology analysis based on the patient's blood pathology area images, specifically:

[0010] Construct an adaptive analysis mechanism to perform pathological analysis from different angles based on image data from different regions, including:

[0011] Pathological analysis of the primary pathological area based on the area of ​​the lesion and the pathological analysis of the secondary pathological area based on the concentration of lesion cells;

[0012] The pathology auxiliary diagnosis module is used to provide medical assistance based on pathology analysis results, specifically:

[0013] Based on the pathological analysis results of the patient by the pathological analysis module, the proportion of the analysis results of each area is calculated, and then a linear analysis function is constructed. Based on the in-depth analysis of the linear analysis function, pathology-assisted diagnosis is achieved.

[0014] As a preferred solution of the pathology-assisted diagnosis system based on image analysis of the present invention, the image data is divided into three regions using a segmentation algorithm as follows:

[0015] For image data I of size H×W, a segmentation algorithm is used to segment the image data into three regions K, including the primary pathological region, the secondary pathological region, and the non-pathological region. The regions are divided according to the probability that each pixel in the image data belongs to each category, specifically:

[0016] For each pixel (i, j) in the image data, calculate the probabilities of the three categories P0(i, j), P1(i, j), P2(i, j), and then select the largest probability P k The category corresponding to (i,j) is,

[0017]

[0018] k∈K∈[0,2]

[0019] Among them, P k (i, j) represents the probability of image data belonging to category k, P0(i, j) represents the probability of image data belonging to category k = 0, P1(i, j) represents the probability of image data belonging to category k = 1, P2(i, j) represents the probability of image data belonging to category k = 2, k max (i, j) represents the probability calculated for pixel (i, j) in the three categories, and the maximum probability category is selected.

[0020] As a preferred solution of the image analysis-based pathology auxiliary diagnosis system of the present invention, the precise region segmentation of the image data is specifically as follows:

[0021] The image data segmentation area is divided according to the calculated maximum probability category, then,

[0022] Setting the regional threshold T1 and the regional threshold T2 to divide the primary pathological area, the secondary pathological area and the non-pathological area;

[0023] The segmentation area is divided according to the set area threshold and the calculated maximum probability category, specifically:

[0024] If the calculated maximum probability category satisfies the formula k max (i, j) = 1, and the probability of pixel (i, j) in k = 1 satisfies the formula T2 > P1(i, j) ≥ T1, indicating that the area where pixel (i, j) is located is a primary pathological area;

[0025] If the calculated maximum probability category satisfies the formula k max (i, j) = 2, and the probability of pixel (i, j) in k = 2 satisfies the formula P2(i, j) ≥ T2, indicating that the area where pixel (i, j) is located is a secondary pathological area;

[0026] If the calculated maximum probability category satisfies the formula k max (i, j) = 0, and the probability of pixel (i, j) in k = 0 at this time satisfies the formula P1(i, j) < T1, indicating that the area where pixel (i, j) is located is a non-pathological area.

[0027] As a preferred solution of the pathology-assisted diagnosis system based on image analysis of the present invention, the conversion of the convolution operation result into the probability of the corresponding category is specifically as follows:

[0028] For the input image data I, the features of the image are gradually extracted through the convolution operation of each layer. The output of each convolution layer is the output of the previous layer after the convolution kernel W. l-1 and the bias term b l-1 The result after the operation is,

[0029] f l =W l-1 ·f l-1 +b l-1

[0030] Among them, W l-1 、b l-1 Represent the convolution kernel and bias term of the l-1 layer respectively, f l Represents the convolution result of the lth layer;

[0031] According to the last convolution result of the network structure, the score of each pixel (i, j) corresponding to category k is calculated, then,

[0032] f k (i,j)=W k ·f l＝4 +b k

[0033] Among them, f k(i, j) represents the score of each pixel (i, j) corresponding to category k, W k 、b k Represent the convolution kernel and bias term of category k respectively;

[0034] Convert the calculated score of each pixel (i, j) corresponding to category k into the corresponding probability, then we have,

[0035]

[0036] And satisfy the formula,

[0037]

[0038] Among them, f k (i, j) represents the score of each pixel (i, j) corresponding to category k, P k (i, j) represents the probability of belonging to category k in the image data. According to the value of k, the segmentation of different regions of the image data is achieved.

[0039] As a preferred solution of the image analysis-based pathology auxiliary diagnosis system of the present invention, the pathological analysis of the primary pathological region according to the area of ​​the lesion region is specifically as follows:

[0040] For the ratio S / S T , set a time period t, after the time period t, count the area of ​​the lesion area S again t And the lesion area threshold S T The ratio S t / S T , and compare the two ratio results, then we have,

[0041] If the comparison results of two consecutive times satisfy the formula (S / S T )<(S t / S T ), indicating that the area of ​​the lesion region in the primary pathological region expands with the extension of the time period, indicating that the pathological degree of the current patient is deepening. For the pathological analysis of the current patient, the secondary pathological region is divided into two levels for analysis;

[0042] If the comparison results of two consecutive times satisfy the formula (S / S T )>(S t / S T ), indicating that the area of ​​the lesion region of the first-level pathological region decreases as the time period increases. For the pathological analysis of the current patient, it is divided into the first-level pathological region for analysis. As the time period increases, if the probability of the pixels in the current region satisfies the formula P1(i,j)<T1, the current patient is divided into the non-pathological region for analysis;

[0043] If the comparison results of two consecutive times satisfy the formula (S / S T )=(S t / S T ), indicating that the area of ​​the lesion region of the first-level pathological region does not change with the extension of the time period, the lesion degree of the current region is continuously monitored, and after the next time period t, the ratio of the current time period to the lesion region area threshold is compared with the ratio of the next time period to the lesion region area threshold. If the lesion region area of ​​the current region still does not change with the extension of the time period, the current patient is kept in the first-level pathological region for analysis.

[0044] As a preferred solution of the image analysis-based pathology auxiliary diagnosis system of the present invention, the pathological analysis of the secondary pathological area according to the concentration of diseased cells is specifically as follows:

[0045] If the calculated proportion of diseased cells in the secondary pathological area η is equal to the set baseline concentration coefficient η T The comparison results satisfy the formula η>η T When , it means that the concentration ratio of pathological cells in the secondary pathological area exceeds the baseline concentration ratio coefficient, indicating that the current patient's pathological degree is deepening. For the current patient's pathological analysis, relevant personnel should be notified immediately to conduct pathological control;

[0046] If the comparison result between the calculated proportion of the diseased cell concentration in the secondary pathological area and the set reference concentration proportion coefficient satisfies the formula η≤η T When , it means that the concentration ratio of pathological cells in the secondary pathological area does not exceed the reference concentration ratio coefficient. According to the number of pathological cells, the pathological degree analysis is performed,

[0047] Continuously monitor the number of diseased cells in the lesion area and extract the number of diseased cells C′ at the current moment I2 And the number of diseased cells C″ after the next cycle t I2 If the number of pathological cells at the current moment is less than the number of pathological cells after the next cycle, it means that the degree of lesions in the secondary pathological area is worsening. For the pathological analysis of the current patient, relevant personnel should be notified immediately to conduct pathological control;

[0048] If the number of pathological cells at the current moment is greater than the number of pathological cells after the next cycle, it means that the degree of lesions in the secondary pathological area is decreasing. For the pathological analysis of the current patient, it is necessary to divide it into the primary pathological area for in-depth analysis;

[0049] If the number of diseased cells at the current moment is balanced with the number of diseased cells after the next cycle, it means that the degree of disease in the secondary pathological area remains unchanged. The pathological analysis of the current patient is maintained in the secondary pathological area.

[0050] As a preferred solution of the pathology-assisted diagnosis system based on image analysis of the present invention, the calculation of the proportion of the analysis results of each region is as follows:

[0051] According to the data analysis results of the pathology analysis module on the patient image, including the non-pathology area analysis results, the first-level pathology area analysis results and the second-level pathology area analysis results, they are marked as R0, R1 and R2 respectively;

[0052] According to the marked analysis results, the analysis results of the same patient are counted and the proportion of the analysis results of each region is calculated. Then,

[0053]

[0054] Among them, p0 represents the proportion of analysis results of non-pathological areas, p1 represents the proportion of analysis results of primary pathological areas, and p2 represents the proportion of analysis results of secondary pathological areas, and the relationship between the three satisfies the formula p0+p1+p2=1.

[0055] As a preferred solution of the pathology-assisted diagnosis system based on image analysis of the present invention, the pathology-assisted diagnosis based on the in-depth analysis of the linear analysis function is specifically implemented as follows:

[0056] According to the proportion of the divided area results, a linear analysis function is constructed. At the same time, a weight coefficient is set for each area, then,

[0057] F=w0·p0+w1·p1+w2·p2

[0058] Among them, w0, w1, and w2 represent the weight coefficients of non-pathological areas, primary pathological areas, and secondary pathological areas, respectively; p0, p1, and p2 represent the proportion of analysis results of non-pathological areas, primary pathological areas, and secondary pathological areas, respectively; and F represents the combined result of the proportions of different areas, which is used for pathological assistance in medical diagnosis, specifically:

[0059] If the combination results of different regional proportions satisfy the formula Indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the current patient's pathological diagnosis is a non-pathological patient;

[0060] If the combination results of different regional proportions satisfy the formula Indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the current patient's pathological diagnosis is a mild patient;

[0061] If the combination results of different regional proportions satisfy the formula It indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the pathological diagnosis of the current patient is a moderate patient.

[0062] A computer device includes a memory and a processor, wherein the memory stores a computer program and the processor implements a pathology auxiliary diagnosis system based on image analysis when executing the computer program.

[0063] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a pathology-assisted diagnosis system based on image analysis.

[0064] Beneficial effects of the present invention:

[0065] The present invention uses a multi-layer convolutional network to extract image features layer by layer, and combines it with a segmentation algorithm to achieve three-category accurate segmentation of pathological areas, breaking through the accuracy limitations of traditional segmentation algorithms in complex images.

[0066] Dynamic thresholds are set for different pathological areas, and a dynamic assessment model for the degree of lesions is constructed by combining time series analysis and cell concentration analysis, which solves the problem that static analysis cannot track the progression of the disease.

[0067] By weightedly integrating the proportions of analysis results of non-pathological, primary pathological, and secondary pathological areas, and constructing a linear analysis function, multi-indicator collaborative diagnosis is achieved, which improves the objectivity and comprehensiveness of the results. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0069] Figure 1 This is a schematic diagram of the overall system structure of the image analysis-based pathology auxiliary diagnosis system of the present invention. DETAILED DESCRIPTION

[0070] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0071] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0072] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0073] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0074] At the same time, in the description of the present invention, it should be noted that the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0075] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0076] Example 1

[0077] Reference Figure 1 , which is an embodiment of the present invention, provides a pathology-assisted diagnosis system based on image analysis, including a data acquisition module for collecting images of the patient's blood pathology area; a pathology analysis module for performing multi-angle pathology analysis based on the patient's blood pathology area images; and a pathology-assisted diagnosis module for providing medical assistance based on the pathology analysis results.

[0078] Specifically, the data acquisition module obtains the image data of the patient's blood pathology area, and uses the segmentation algorithm to divide the acquired image data into multiple areas in order to analyze different degrees of pathology and non-pathology areas. According to the areas divided by the data acquisition module, the pathology analysis module performs multi-angle pathology analysis on the images of the pathology areas with different degrees, and sends them to the pathology auxiliary diagnosis module. The latter performs pathology assistance for medical diagnosis based on the received medical record analysis results.

[0079] Furthermore, the acquisition of the patient's pathological region image data is performed by using a high-precision pathology scanner to acquire the patient's pathological image data I, and setting the image size thereof to be H×W;

[0080] The method of dividing the acquired image data into multiple regions using a segmentation algorithm is a segmentation algorithm based on deep learning, which divides the acquired image data into pathological regions of different degrees. The specific implementation is as follows:

[0081] For image data I of size H×W, a segmentation algorithm is used to segment the image data into three regions K, including the primary pathological region, the secondary pathological region, and the non-pathological region. The regions are divided according to the probability that each pixel in the image data belongs to each category, specifically:

[0082] For each pixel (i, j) in the image data, calculate the probabilities of the three categories P0(i, j), P1(i, j), P2(i, j), and then select the largest probability P k The category corresponding to (i,j) is,

[0083]

[0084] k∈K∈[0,2]

[0085] Among them, P k (i, j) represents the probability of image data belonging to category k, P0(i, j) represents the probability of image data belonging to category k = 0, P1(i, j) represents the probability of image data belonging to category k = 1, P2(i, j) represents the probability of image data belonging to category k = 2, k max (i, j) represents the probability calculated for pixel (i, j) in the three categories, and the maximum probability category is selected;

[0086] The image data segmentation area is divided according to the calculated maximum probability category, then,

[0087] Setting the regional threshold T1 and the regional threshold T2 to divide the primary pathological area, the secondary pathological area and the non-pathological area;

[0088] The segmentation area is divided according to the set area threshold and the calculated maximum probability category, specifically:

[0089] If the calculated maximum probability category satisfies the formula k max (i, j) = 1, and the probability of pixel (i, j) in k = 1 satisfies the formula T2 > P1(i, j) ≥ T1, indicating that the area where pixel (i, j) is located is a primary pathological area;

[0090] If the calculated maximum probability category satisfies the formula k max (i, j) = 2, and the probability of pixel (i, j) in k = 2 satisfies the formula P2(i, j) ≥ T2, indicating that the area where pixel (i, j) is located is a secondary pathological area;

[0091] If the calculated maximum probability category satisfies the formula k max (i, j) = 0, and the probability of pixel (i, j) in k = 0 at this time satisfies the formula P1(i, j) < T1, indicating that the area where pixel (i, j) is located is a non-pathological area.

[0092] It should be noted that the probability of pixel (i, j) in each category is calculated based on the deep learning segmentation algorithm, as follows:

[0093] Based on the number of divided categories, a network structure with the same number of layers as the number of categories is constructed. According to the constructed network structure, convolution operations are performed layer by layer, specifically:

[0094] For the input image data I, the features of the image are gradually extracted through the convolution operation of each layer. The output of each convolution layer is the output of the previous layer after the convolution kernel W. l-1 and the bias term b l-1 The result after the operation is,

[0095] f l =W l-1 ·f l-1 +b l-1

[0096] Among them, W l-1 、b l-1 They represent the convolution kernel and bias term of the l-1th layer respectively, which are set by the implementer according to the actual application scenario. l represents the convolution result of the lth layer. In the present invention, when l=3, it is the convolution result of the last layer;

[0097] According to the last convolution result of the network structure, the score of each pixel (i, j) corresponding to category k is calculated, then,

[0098] f k (i,j)=W k ·f l＝4 +b k

[0099] Among them, f k (i, j) represents the score of each pixel (i, j) corresponding to category k, W k 、b kThey represent the convolution kernel and bias term of category k respectively, which are set by the implementer according to the actual application scenario;

[0100] Convert the calculated score of each pixel (i, j) corresponding to category k into the corresponding probability, then we have,

[0101]

[0102] And satisfy the formula,

[0103]

[0104] Among them, f k (i, j) represents the score of each pixel (i, j) corresponding to category k, P k (i, j) represents the probability of belonging to category k in the image data. According to the value of k, the segmentation of different regions of the image data is achieved.

[0105] Furthermore, the multi-angle pathological analysis is performed according to the images of different pathological regions with different degrees by constructing an adaptive analysis mechanism. Different analysis angles are adaptively selected for pathological analysis of image data in different regions, and finally the analysis results of all angles are combined into the pathological analysis results of the image data. The specific analysis is as follows:

[0106] The adaptive analysis mechanism analyzes the pathology degree from different perspectives based on the regions divided by the image data, specifically:

[0107] For non-pathological areas in the input image data, the current area is a normal area, and real-time monitoring is maintained for healthy tissues in the current area. If the probability formula of the pixels in the current area satisfies P1(i,j)≥T1, it means that the pathology of the current area has deepened, and the analysis rules of the current area are divided into the first-level pathological area for analysis;

[0108] For the first-level pathological area in the input image data, which is a mild lesion, the pathological analysis of the current area is realized based on the shape analysis of the pathological area. Then,

[0109] The lesion area threshold S of the first-level pathological area is set based on the lesion data of the first-level pathological area in the historical database. T At the same time, according to the pixels of the current region (first-level pathological region), the area of ​​the lesion region in the current region is calculated, then,

[0110] Determine the lesion area A, and calculate the area of ​​the lesion area in the current area according to the pixels (i, j) of the current area (first-level pathological area). Then,

[0111]

[0112] Where S represents the area of ​​the lesion region in the current region (the first-level pathological region), H1 represents the height of the first-level pathological region, W1 represents the width of the first-level pathological region, and B(i, j) represents whether the pixel (i, j) in the current region is within the lesion region A. The specific judgment is as follows:

[0113] If the pixel (i, j) in the current area is in the lesion area A, the value of B(i, j) is 1;

[0114] If the pixel (i, j) in the current area is not in the lesion area A, the value of B(i, j) is 0;

[0115] According to the calculated lesion area S and lesion area threshold S T The ratio between them is used to realize the pathological analysis of the current area (first-level lesion area), specifically:

[0116] For the ratio S / S T , set a time period t, after the time period t, count the area of ​​the lesion area S again t And the lesion area threshold S T The ratio S t / S T , and compare the two ratio results, then we have,

[0117] If the comparison results of two consecutive times satisfy the formula (S / S T )<(S t / S T ), indicating that the area of ​​the lesion region in the current region (the primary pathological region) expands with the extension of the time period, indicating that the pathological degree of the current patient is deepening. For the pathological analysis of the current patient, the secondary pathological region is divided into two levels for analysis;

[0118] If the comparison results of two consecutive times satisfy the formula (S / S T )>(S t / S T ), indicating that the area of ​​the lesion region in the current region (the first-level pathological region) decreases as the time period increases. For the pathological analysis of the current patient, the patient is divided into the first-level pathological region for analysis. As the time period increases, if the probability of the pixels in the current region satisfies the formula P1(i, j) < T1, the patient is divided into the non-pathological region for analysis.

[0119] If the comparison results of two consecutive times satisfy the formula (S / S T )=(S t / S T), indicating that the area of ​​the lesion region in the current region (the first-level pathological region) does not change with the extension of the time period, the lesion degree in the current region is continuously monitored, and after the next time period t, the ratio of the current time period to the lesion region area threshold is compared with the ratio of the next time period to the lesion region area threshold. If the lesion region area in the current region still does not change with the extension of the time period, the current patient is kept in the first-level pathological region for analysis.

[0120] For the secondary pathological area in the input image data, which is a moderate lesion, the pathological analysis of the current area is realized based on the texture depth analysis of the pathological area. Then,

[0121] The lesion area S2 in the secondary pathological region is calculated according to the calculation method of the lesion area in the primary pathological region. At the same time, the number of lesion cells C in the secondary pathological region is counted. I2 and the number of diseased cells in the lesion area C′ I2 ;

[0122] According to the statistical number of pathological cells, the concentration of pathological cells in the secondary pathological area and the concentration of pathological cells in the pathological area are calculated respectively, and then,

[0123] The concentration of diseased cells in the secondary pathological area,

[0124]

[0125] The concentration of diseased cells in the lesion area,

[0126]

[0127] Among them, C I2 represents the number of diseased cells in the secondary pathological area, C′ I2 represents the number of pathological cells in the lesion area, A2 represents the area of ​​the secondary pathological area, S2 represents the area of ​​the lesion area, and c I2 represents the concentration of diseased cells in the secondary pathological area, c′ I2 It indicates the concentration of diseased cells in the lesion area;

[0128] According to the calculated cell concentration, the proportion of diseased cells is determined, then,

[0129]

[0130] Among them, c I2 represents the concentration of diseased cells in the secondary pathological area, c′ I2 represents the concentration of pathological cells in the lesion area, and η represents the proportion of pathological cell concentration in the secondary pathological area, which is used to analyze the pathological degree of the secondary pathological area. Specifically:

[0131] Before analyzing the pathological degree of the secondary pathological area, set the baseline concentration ratio coefficient η of the pathological cells in the pathological area. T , according to the set baseline concentration ratio coefficient, the pathological degree analysis is carried out,

[0132] If the comparison result between the calculated proportion of the diseased cell concentration in the secondary pathological area and the set reference concentration proportion coefficient satisfies the formula η>η T When , it means that the concentration ratio of pathological cells in the current area (secondary pathological area) exceeds the baseline concentration ratio coefficient, indicating that the pathological degree of the current patient is deepening. For the pathological analysis of the current patient, relevant personnel will be notified immediately to conduct pathological control;

[0133] If the comparison result between the calculated proportion of the diseased cell concentration in the secondary pathological area and the set reference concentration proportion coefficient satisfies the formula η≤η T When , it means that the proportion of pathological cell concentration in the current area (secondary pathological area) does not exceed the reference concentration proportion coefficient. According to the number of pathological cells, the pathological degree analysis is performed,

[0134] Continuously monitor the number of diseased cells in the lesion area and extract the number of diseased cells C′ at the current moment I2 And the number of diseased cells C″ after the next cycle t I2 If the number of pathological cells at the current moment is less than the number of pathological cells after the next cycle, it means that the degree of lesions in the current area (secondary pathological area) is worsening. For the pathological analysis of the current patient, relevant personnel are immediately notified to conduct pathological control;

[0135] If the number of pathological cells at the current moment is greater than the number of pathological cells after the next cycle, it means that the degree of lesions in the current area (secondary pathological area) is decreasing. For the pathological analysis of the current patient, it is necessary to divide it into the primary pathological area for in-depth analysis;

[0136] If the number of diseased cells at the current moment is balanced with the number of diseased cells after the next cycle, it means that the degree of lesions in the current area (secondary pathological area) remains unchanged. For the pathological analysis of the current patient, the analysis is maintained in the secondary pathological area.

[0137] Furthermore, the pathology-assisted medical diagnosis is based on the analysis results of the pathology analysis module for the patient. By combining the analysis results of different regions and performing in-depth analysis on the combined results, pathology-assisted medical diagnosis is achieved. The specific implementation is as follows:

[0138] According to the data analysis results of the pathology analysis module on the patient image, including the non-pathology area analysis results, the first-level pathology area analysis results and the second-level pathology area analysis results, they are marked as R0, R1 and R2 respectively;

[0139] According to the marked analysis results, the analysis results of the same patient are counted and the proportion of the analysis results of each region is calculated. Then,

[0140]

[0141] Wherein, p0 represents the proportion of analysis results of non-pathological areas, p1 represents the proportion of analysis results of primary pathological areas, and p2 represents the proportion of analysis results of secondary pathological areas, and the relationship between the three satisfies the formula p0+p1+p2=1;

[0142] According to the proportion of the divided regional results, a linear analysis function is constructed. At the same time, the weight coefficient is set for each region to limit the technical solution. Then,

[0143] F=w0·p0+w1·p1+w2·p2

[0144] Among them, w0, w0, w0 represent the weight coefficients of non-pathological area, primary pathological area and secondary pathological area respectively, which are set by implementers according to actual application scenarios. p0, p1, p2 represent the proportion of analysis results of non-pathological area, primary pathological area and secondary pathological area respectively. F represents the combined result of the proportions of different areas, which is used for pathological assistance in medical diagnosis. Specifically:

[0145] If the combination results of different regional proportions satisfy the formula Indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the current patient's pathological diagnosis is a non-pathological patient;

[0146] If the combination results of different regional proportions satisfy the formula Indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the current patient's pathological diagnosis is a mild patient;

[0147] If the combination results of different regional proportions satisfy the formula It indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the pathological diagnosis of the current patient is a moderate patient.

[0148] Furthermore, if the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the system described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0149] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0150] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0151] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those that are not relevant to implementing the invention) may not be described.

[0152] It will be appreciated that in the development of any actual embodiment, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but will, for those of ordinary skill having the benefit of this disclosure, be a routine undertaking of design, fabrication, and production without undue experimentation.

[0153] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A pathology-assisted diagnosis system based on image analysis, characterized by: include, The data acquisition module is used to collect images of the patient's blood pathology area, specifically: By collecting the patient's pathological image data and using the segmentation algorithm to divide the image data into three regions, the image data is accurately segmented according to the probability of the pixel points in each category, including: By constructing a multi-layer convolutional network structure, convolution operations are performed layer by layer, and finally the convolution operation results are converted into the probability of the corresponding category; The pathology analysis module is used to perform multi-angle pathology analysis based on the patient's blood pathology area images, specifically: Construct an adaptive analysis mechanism to perform pathological analysis from different angles based on image data from different regions, including: Pathological analysis of the primary pathological area based on the area of ​​the lesion and the pathological analysis of the secondary pathological area based on the concentration of lesion cells; The pathology auxiliary diagnosis module is used to provide medical assistance for pathology analysis results, specifically: Based on the pathological analysis results of the patient by the pathological analysis module, the proportion of the analysis results of each area is calculated, and then a linear analysis function is constructed. Based on the in-depth analysis of the linear analysis function, pathology-assisted diagnosis is achieved.

2. The pathology-assisted diagnosis system based on image analysis according to claim 1, characterized in that: The image data is divided into three regions using a segmentation algorithm as follows: For image data I of size H×W, a segmentation algorithm is used to segment the image data into three regions K, including the primary pathological region, the secondary pathological region, and the non-pathological region. The regions are divided according to the probability that each pixel in the image data belongs to each category, specifically: For each pixel (i, j) in the image data, calculate the probabilities of the three categories P0(i, j), P1(i, j), P2(i, j), and then select the largest probability P k The category corresponding to (i,j) is, Among them, P k (i, j) represents the probability of image data belonging to category k, P0(i, j) represents the probability of image data belonging to category k = 0, P1(i, j) represents the probability of image data belonging to category k = 1, P2(i, j) represents the probability of image data belonging to category k = 2, k max (i, j) represents the probability calculated for pixel (i, j) in the three categories, and the maximum probability category is selected.

3. The pathology-assisted diagnosis system based on image analysis according to claim 2, characterized in that: The precise region segmentation of the image data is specifically as follows: The image data segmentation area is divided according to the calculated maximum probability category, then, Setting the regional threshold T1 and the regional threshold T2 to divide the primary pathological area, the secondary pathological area and the non-pathological area; The segmentation area is divided according to the set area threshold and the calculated maximum probability category, specifically: If the calculated maximum probability category satisfies the formula k max (i, j) = 1, and the probability of pixel (i, j) in k = 1 satisfies the formula T2 > P1(i, j) ≥ T1, indicating that the area where pixel (i, j) is located is a primary pathological area; If the calculated maximum probability category satisfies the formula k max (i, j) = 2, and the probability of pixel (i, j) in k = 2 satisfies the formula P2(i, j) ≥ T2, indicating that the area where pixel (i, j) is located is a secondary pathological area; If the calculated maximum probability category satisfies the formula k max (i, j) = 0, and the probability of pixel (i, j) in k = 0 at this time satisfies the formula P1(i, j) < T1, indicating that the area where pixel (i, j) is located is a non-pathological area.

4. The pathology-assisted diagnosis system based on image analysis according to claim 3, characterized in that: The specific conversion of the convolution operation result into the probability of the corresponding category is as follows: For the input image data I, the features of the image are gradually extracted through the convolution operation of each layer. The output of each convolution layer is the output of the previous layer after the convolution kernel W. l-1 and the bias term b l-1 The result after the operation is, f l =W l-1 ·f l-1 +b l-1 Among them, W l-1 、b l-1 Represent the convolution kernel and bias term of the l-1 layer respectively, f l Represents the convolution result of the lth layer; According to the last convolution result of the network structure, the score of each pixel (i, j) corresponding to category k is calculated, then, f k (i,j)=W k ·f l＝4 +b k Among them, f k (i, j) represents the score of each pixel (i, j) corresponding to category k, W k 、b k Represent the convolution kernel and bias term of category k respectively; Convert the calculated score of each pixel (i, j) corresponding to category k into the corresponding probability, then we have, And satisfy the formula, Among them, f k (i, j) represents the score of each pixel (i, j) corresponding to category k, P k (i, j) represents the probability of belonging to category k in the image data. According to the value of k, the segmentation of different regions of the image data is achieved.

5. The pathology-assisted diagnosis system based on image analysis according to claim 4, characterized in that: The pathological analysis of the primary pathological region according to the area of ​​the lesion region is as follows: For the ratio S / S T , set a time period t, after the time period t, count the area of ​​the lesion area S again t And the lesion area threshold S T The ratio S t / S T , and compare the two ratio results, then we have, If the comparison results of two consecutive times satisfy the formula (S / S T )<(S t / S T ), indicating that the area of ​​the lesion region in the primary pathological region expands with the extension of the time period, indicating that the pathological degree of the current patient is deepening. For the pathological analysis of the current patient, the secondary pathological region is divided into two levels for analysis; If the comparison results of two consecutive times satisfy the formula (S / S T )>(S t / S T ), indicating that the area of ​​the lesion region of the first-level pathological region decreases as the time period increases. For the pathological analysis of the current patient, it is divided into the first-level pathological region for analysis. As the time period increases, if the probability of the pixels in the current region satisfies the formula P1(i,j)<T1, the current patient is divided into the non-pathological region for analysis; If the comparison results of two consecutive times satisfy the formula (S / S T )=(S t / S T ), indicating that the area of ​​the lesion region of the first-level pathological region does not change with the extension of the time period, the lesion degree of the current region is continuously monitored, and after the next time period t, the ratio of the current time period to the lesion region area threshold is compared with the ratio of the next time period to the lesion region area threshold. If the lesion region area of ​​the current region still does not change with the extension of the time period, the current patient is kept in the first-level pathological region for analysis.

6. The pathology-assisted diagnosis system based on image analysis according to claim 5, characterized in that: The pathological analysis of the secondary pathological area according to the concentration of diseased cells is as follows: If the calculated proportion of diseased cells in the secondary pathological area η is equal to the set baseline concentration coefficient η T The comparison results satisfy the formula η>η T When , it means that the concentration ratio of pathological cells in the secondary pathological area exceeds the baseline concentration ratio coefficient, indicating that the current patient's pathological degree is deepening. For the current patient's pathological analysis, relevant personnel should be notified immediately to conduct pathological control; If the comparison result between the calculated proportion of the diseased cell concentration in the secondary pathological area and the set reference concentration proportion coefficient satisfies the formula η≤η T When , it means that the concentration ratio of pathological cells in the secondary pathological area does not exceed the reference concentration ratio coefficient. According to the number of pathological cells, the pathological degree analysis is performed, Continuously monitor the number of diseased cells in the lesion area and extract the number of diseased cells C′ at the current moment I2 And the number of diseased cells C″ after the next cycle t I2 If the number of pathological cells at the current moment is less than the number of pathological cells after the next cycle, it means that the degree of lesions in the secondary pathological area is worsening. For the pathological analysis of the current patient, relevant personnel should be notified immediately to conduct pathological control; If the number of pathological cells at the current moment is greater than the number of pathological cells after the next cycle, it means that the degree of lesions in the secondary pathological area is decreasing. For the pathological analysis of the current patient, it is necessary to divide it into the primary pathological area for in-depth analysis; If the number of diseased cells at the current moment is balanced with the number of diseased cells after the next cycle, it means that the degree of disease in the secondary pathological area remains unchanged. The pathological analysis of the current patient is maintained in the secondary pathological area.

7. The pathology-assisted diagnosis system based on image analysis according to claim 6, characterized in that: The specific calculation of the analysis results of each region is as follows: According to the data analysis results of the pathology analysis module on the patient image, including the non-pathology area analysis results, the first-level pathology area analysis results and the second-level pathology area analysis results, they are marked as R0, R1 and R2 respectively; According to the marked analysis results, the analysis results of the same patient are counted and the proportion of the analysis results of each region is calculated. Then, Among them, p0 represents the proportion of analysis results of non-pathological areas, p1 represents the proportion of analysis results of primary pathological areas, and p2 represents the proportion of analysis results of secondary pathological areas, and the relationship between the three satisfies the formula p0+p1+p2=1.

8. The pathology-assisted diagnosis system based on image analysis according to claim 7, characterized in that: The detailed description of the pathology-assisted diagnosis based on the in-depth analysis of the linear analysis function is as follows: According to the proportion of the divided area results, a linear analysis function is constructed. At the same time, a weight coefficient is set for each area, then, F=w0·p0+w1·p1+w2·p2 Among them, w0, w1, and w2 represent the weight coefficients of non-pathological areas, primary pathological areas, and secondary pathological areas, respectively; p0, p1, and p2 represent the proportion of analysis results of non-pathological areas, primary pathological areas, and secondary pathological areas, respectively; and F represents the combined result of the proportions of different areas, which is used for pathological assistance in medical diagnosis, specifically: If the combination results of different regional proportions satisfy the formula Indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the current patient's pathological diagnosis is a non-pathological patient; If the combination results of different regional proportions satisfy the formula Indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the current patient's pathological diagnosis is a mild patient; If the combination results of different regional proportions satisfy the formula It indicates that the non-pathological area accounts for a high proportion in the pathological analysis results of the current patient, and the pathological diagnosis of the current patient is a moderate patient.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the system according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the system according to any one of claims 1 to 8 is implemented.

Images