Pathological section diagnosis area marking and archiving method

Abstract

The invention discloses a pathological section diagnosis area marking and archiving method, which comprises the following steps: acquiring multi-temporal pathological section image data of a target tissue area, the multi-temporal pathological section image data comprising section images of the target tissue area collected at different time points; based on the multi-temporal pathological section image data, constructing a cable pipeline topology network in the target organization area through an image registration algorithm and a topology refinement algorithm; for the pathological section image data of each time phase, identifying an abnormal diagnosis area in the pathological section image data through an image identification algorithm; calculating geometric topological relation data between the spatial position of the abnormal diagnosis area identified in each time phase and the nearest branch node in the cable pipeline topological network; and based on the geometric topological relation data, generating fault source coordinate data in one-to-one correspondence with the abnormal diagnosis areas, and storing the fault source coordinate data into a diagnosis area coordinate database.

Description

Technical Field

[0001] This application relates to the technical field of pathological slide marking, and more particularly to a method for archiving diagnostic area markings on pathological slides. Background Technology

[0002] In pathological diagnosis and tumor progression research, precise spatial localization and dynamic tracking of lesion areas (such as tumor foci) are crucial for understanding their occurrence and development mechanisms. Traditional methods of pathological slide analysis and digital archiving typically involve marking isolated lesion regions (ROIs) on the entire slide image, recording their absolute coordinates or relative positions with respect to the slide boundaries. However, the biological behavior of lesions such as tumors, especially their invasion and metastasis, is often not random but tends to follow the anatomical pathways of the tissue's inherent vascular system (such as blood vessels and lymphatic vessels).

[0003] Traditional labeling methods primarily rely on the mechanical coordinates of the scanner stage or the pixel coordinates of the image itself. However, due to factors such as the embedding angle of tissue after ex vivo, tissue shrinkage during slide preparation, and differences in placement at different acquisition times, precise alignment of lesion locations across time phases within the same tissue cannot be achieved on the coordinate axes. Tumor progression (such as invasion and spread) often follows biological patterns, frequently spreading along blood vessels, lymphatic vessels, or the vascular system. Existing labeling methods mostly target isolated lesions, lacking records of the topological relationships between lesions and surrounding anatomical structures, making it difficult to reproduce the true path of tumor spread in research. When comparing the progression of lesions in the same patient at different stages (e.g., evaluating the effectiveness of chemotherapy), physicians can only rely on vague comparisons based on visual memory, lacking quantitative, structured data to support the determination of the direction and speed of progression, resulting in low secondary utilization value of archived data in research analysis. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] To achieve the above objectives, a first aspect of this application proposes a method for archiving diagnostic region markings on pathological slides, the method being executed by a processor and including the following steps:

[0006] Acquire multi-temporal pathological slide image data of the target tissue region, wherein the multi-temporal pathological slide image data includes slide images of the target tissue region collected at different time points;

[0007] Based on the multi-temporal pathological slice image data, a cable conduit topology network is constructed within the target tissue area using an image registration algorithm and a topology refinement algorithm. The cable conduit topology network includes a cable conduit skeleton characterized by a centerline and multiple branch nodes located on the cable conduit skeleton.

[0008] For the pathological slide image data of each time phase, abnormal diagnostic areas in the pathological slide image data are identified by an image recognition algorithm;

[0009] For each time phase, the identified abnormal diagnostic area is used to calculate the geometric topological relationship data between its spatial location and the nearest neighboring branch node in the cable duct topology network. The geometric topological relationship data includes distance value and relative azimuth angle.

[0010] Based on the geometric topology data, fault source coordinate data corresponding one-to-one with the abnormal diagnosis area is generated. The fault source coordinate data uses the branch nodes in the cable duct topology network as spatial reference benchmarks and is stored in the diagnosis area coordinate database.

[0011] In addition, the pathological slide diagnostic area marking and archiving method proposed in this application may also have the following additional technical features:

[0012] In one embodiment of this application, the steps for constructing the cable duct topology network are as follows:

[0013] The multi-temporal pathological slice image data is segmented to extract cable and pipe structure images from the target tissue region;

[0014] The cable duct structure image is processed using a topology thinning algorithm to obtain a cable duct skeleton with a single pixel width;

[0015] Identify pixels in the cable conduit skeleton with a connection count greater than 2 and determine them as branch nodes.

[0016] In one embodiment of this application, the method for identifying the abnormal diagnostic region is as follows:

[0017] Using an image segmentation algorithm based on gradient or intensity threshold, connected regions that are significantly different from the surrounding tissue are segmented from the pathological slide image data, and these connected regions are marked as the abnormal diagnostic regions.

[0018] In one embodiment of this application, the relative azimuth angle is calculated as follows:

[0019] Taking the nearest neighboring branch node as the origin and the tangent direction of the skeleton segment where the node is located as the reference axis, calculate the angle between the direction vector of the centroid of the abnormal diagnosis region relative to the origin and the reference axis, and determine the angle as the relative azimuth angle.

[0020] In one embodiment of this application, the distance value is calculated as follows:

[0021] Calculate the Euclidean geometric distance from the centroid of the abnormal diagnostic region to its nearest branch node, and determine the Euclidean geometric distance as the distance value.

[0022] In one embodiment of this application, storing the fault source coordinate data into the diagnostic area coordinate database includes:

[0023] The fault source coordinate data, the corresponding time phase information, and the corresponding target organization area identifier are associated and stored to form a structured database record.

[0024] In one embodiment of this application, the specific steps for obtaining the connected region are as follows:

[0025] The pathological slide image data is converted into first grayscale image data;

[0026] Apply a gradient-based edge detection algorithm or a pixel intensity threshold-based binarization algorithm to the first grayscale image data to generate first binarized image data. The first pixel value in the first binarized image data represents a candidate region that differs from the surrounding tissue, and the second pixel value represents the background region.

[0027] Perform connected component analysis on the first binarized image data to obtain at least one candidate connected component data; based on a preset minimum area threshold, select candidate connected components with an area greater than the minimum area threshold from the at least one candidate connected component data, and mark the selected candidate connected component data as the abnormal diagnosis region data.

[0028] In one embodiment of this application, the method further includes:

[0029] The fault source coordinate data generated at different times for the same target organization region are correlated in time to form the evolution sequence data of the abnormal diagnosis region;

[0030] Based on the changes in the coordinate data of the fault source in adjacent time phases of the evolution sequence data, the evolution direction and evolution speed of the abnormal diagnosis region are calculated, and an evolution trend vector is generated.

[0031] In response to a user query, the system retrieves the coordinate data records of the fault source that conform to a specific spatial evolution pattern from the diagnostic area coordinate database and generates a visual analysis report.

[0032] The pathological slide diagnostic region marking and archiving method of this application utilizes the inherent vascular topology network within the tissue as a stable spatial reference. By calculating the distance and azimuth of the lesion relative to branch nodes, anatomical reference coordinates are generated, effectively overcoming slide deformation and achieving precise alignment across time phases. Based on this, an evolution trend vector is automatically generated and visualized, significantly improving diagnostic efficiency. Simultaneously, massive amounts of image data are transformed into lightweight structured coordinate records, greatly optimizing storage management.

[0033] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0034] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0035] Figure 1 A flowchart illustrating the method for marking and archiving diagnostic areas in pathological slides in this application;

[0036] Figure 2 A flowchart illustrating the construction process of the cable conduit topology network in this application;

[0037] Figure 3 This is a flowchart illustrating the specific steps for obtaining the connected regions in this application;

[0038] Figure 4 This is a flowchart illustrating the calculation of the geometric topological relationships in this application. Detailed Implementation

[0039] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0040] The method for marking and archiving diagnostic areas of pathological slides according to embodiments of this application will be described below with reference to the accompanying drawings.

[0041] like Figures 1-4 As shown in the embodiment of this application, the method for marking and archiving diagnostic regions of pathological slides, executed by a processor, may include the following steps:

[0042] Acquire multi-temporal pathological slide image data of the target tissue region, wherein the multi-temporal pathological slide image data includes slide images of the target tissue region collected at different time points; it should be noted that the multi-temporal pathological slide image data should come from consecutive biopsy surgical samples of the same patient, and high-resolution whole slide digital images formed after digital scanning.

[0043] It should be noted that the system receives raw, high-resolution pathological slide images from different time points (t1, t2, ...). These images are represented as a three-dimensional data matrix in the computer. Where (x, y) are pixel coordinates. Representing the temporal phase. The processor first loads these images into memory, then automatically performs image registration (using traditional algorithms such as SIFT) and normalization operations, outputting a series of spatially aligned and color-normalized image tensors.

[0044] The formula for reference is: , where, by minimizing Function, constantly adjusted Parameters; Used to register the image (Moving image) mapped to reference image In the coordinate system of (fixed image).

[0045] Specifically, the system selects a time phase (such as the earliest t1) as a reference benchmark, calculates the spatial transformation relationship between other time phase images and this benchmark (e.g., how many degrees of rotation or how many pixels of translation are needed), and corrects all images to the same spatial coordinate system. This ensures that the same anatomical structure appears in approximately the same location in images from different time points. Based on the multi-time phase pathological slide image data, a cable conduit topology network is constructed within the target tissue region using image registration and topology refinement algorithms. The cable conduit topology network includes a cable conduit skeleton characterized by a centerline and multiple branch nodes located on the cable conduit skeleton.

[0046] It should be noted that the aforementioned cable conduits may represent anatomical structures in biological tissues that have tubular or linear morphology, such as blood vessels, lymphatic vessels, and nerve bundles.

[0047] The steps for constructing the cable conduit topology network are as follows: segmenting the multi-temporal pathological slice image data and extracting the cable conduit structure image in the target tissue region;

[0048] It should be noted that when constructing the cable duct topology network, the image can be extracted using the U-Net model. U-Net extracts features through the encoder, restores the resolution through the decoder, and outputs a predicted image of the probability that a pixel belongs to a duct.

[0049] ,in, This is a binarized image of the cable conduit; pixel 1 represents the extracted cable conduit structure, and 0 represents the background. Automatic segmentation threshold;

[0050] The cable duct structure image is processed using a topology thinning algorithm to obtain a cable duct skeleton with a single pixel width; pixels with a connection number greater than 2 in the cable duct skeleton are identified and determined as branch nodes.

[0051] It should be noted that the thinning algorithm described in the above embodiments can be the Zhang-Suen thinning algorithm, which strips boundary pixels through iterative scanning without changing the topological connectivity of the object.

[0052] Specifically, corresponding pixels Its 8 fields During the iteration, the following topological constraints must be satisfied simultaneously for a pixel to be deleted:

[0053] (1) Non-isolated and non-endpoint constraint:

[0054] (2) Calculation of connectivity preservation constraints Number of transitions from 0 to 1 within the neighborhood

[0055] The single-pixel lines processed by the above process constitute a cable duct skeleton image with a single-pixel width, and at the same time, provide a unique coordinate starting point for subsequent distance calculations.

[0056] It should be noted that the method for determining branch nodes in the above embodiments is as follows:

[0057] Scan the input cable duct skeleton image with a width of one pixel and calculate its connectivity properties;

[0058] The formulas involved are: ;in, Representing pixels The number of connected components; and . If a pixel connects to more than two directions, it is identified as a branch node. All branch nodes are then merged to obtain the coordinate set of all branch nodes. .

[0059] For the pathological slide image data of each time phase, abnormal diagnostic areas in the pathological slide image data are identified by an image recognition algorithm;

[0060] The method for identifying the abnormal diagnostic region is as follows:

[0061] Using an image segmentation algorithm based on gradient or intensity threshold, connected regions that are significantly different from the surrounding tissue are segmented from the pathological slide image data, and these connected regions are marked as the abnormal diagnostic regions.

[0062] Specifically, the steps for obtaining the connected region are as follows:

[0063] The pathological slide image data is converted into a first grayscale image data. ;

[0064] It should be noted that a weighted average method can be used for conversion. By adjusting the weights, brightness characteristics that conform to visual perception can be preserved. Since this technology is existing, it will not be elaborated on here.

[0065] For the first grayscale image The data is processed using either gradient-based edge detection algorithms or pixel intensity threshold-based binarization algorithms to generate the first binarized image data. In the first binarized image data, the first pixel value represents a candidate region that differs from the surrounding tissue, and the second pixel value represents the background region.

[0066] ,in The globally optimal threshold can be calculated using Otsu's method to extract regions with significant differences.

[0067] Perform connectivity analysis on the first binarized image data to obtain at least one candidate connected region data;

[0068] It should be noted that, in the above embodiments, the connected component analysis of the first binarized image data is performed by using the processor to aggregate adjacent pixels with a pixel value of 1 into a candidate connected component set using the 8-neighborhood connectivity criterion. Each obtained at this time It represents a candidate anomaly target that is morphologically continuous.

[0069] Based on a preset minimum area threshold, candidate connected regions with an area greater than the minimum area threshold are selected from the at least one candidate connected region data, and the selected candidate connected region data is marked as the abnormal diagnosis region data.

[0070] It should be noted that the minimum area threshold in the above embodiments is set based on parameters dynamically determined according to clinical experience and image resolution. For example, if the minimum lesion diameter is known to be approximately... Then convert it to pixel area as .

[0071] The criteria for judgment are If the conditions are met, the region is marked as an abnormal diagnostic area; otherwise, it is removed.

[0072] For each phase of the identified abnormal diagnostic region, the geometric topological relationship data between its spatial location and the nearest neighboring branch node in the cable duct topology network is calculated. The geometric topological relationship data includes distance value and relative azimuth angle. It should be noted that the centroid can be calculated using a method based on image moments.

[0073] The distance value is calculated as follows:

[0074] Calculate the abnormal diagnostic region The centroid (after the above steps of screening) To its nearest branch node The Euclidean geometric distance is used to determine the distance value.

[0075] The calculation formula is: ;at this time This represents the distance between the lesion and the nearest tissue intersection.

[0076] The relative azimuth angle is calculated as follows: taking the nearest neighboring branch node as the origin and the tangent direction of the skeleton segment where the node is located as the reference axis, the angle between the direction vector of the centroid of the abnormal diagnosis area relative to the origin and the reference axis is calculated, and the angle is determined as the relative azimuth angle.

[0077] Calculation formula ;in The direction vector from the node to the centroid;

[0078] ;in, Used as a reference axis; This is the relative azimuth angle, representing the angle of deviation of the lesion from the direction of vascular growth. Even if the image as a whole is rotated by 90°, due to the reference axis... It will also change accordingly, therefore The value remains unchanged.

[0079] Based on the geometric topology data, fault source coordinate data corresponding one-to-one with the abnormal diagnosis area is generated. The fault source coordinate data uses the branch nodes in the cable duct topology network as spatial reference benchmarks and is stored in the diagnosis area coordinate database.

[0080] Storing the fault source coordinate data into the diagnostic area coordinate database includes:

[0081] The fault source coordinate data, the corresponding time phase information, and the corresponding target organization area identifier are associated and stored to form a structured database record.

[0082] It should be noted that the fault source coordinate data obtained in the above embodiments is... ,in, The index corresponding to the minimum value in the list of distance values;

[0083] It should be noted that, This represents a specific anatomical location within human tissue (e.g., the first bifurcation of the left coronary artery). Although the lesion may move or grow at different times, its coordinates will remain the same as long as it is near this branching node. Next, patient information needs to be imported, such as the patient's ID number, organ location, and medical record number. This refers to the label or barcode information for these data.

[0084] Finally, the structured data bits of the fault source coordinates were obtained. .

[0085] Furthermore, the above-mentioned method for marking and archiving diagnostic areas in pathological sections also includes: The fault source coordinate data generated at different times for the same target organization region are correlated in time to form the evolution sequence data of the abnormal diagnosis region;

[0086] It should be noted that the above evolution sequence data is arranged in chronological order, and fully records the displacement history of the lesion relative to the pipeline anchor point.

[0087] Based on the changes in the coordinate data of the fault source in adjacent time phases of the evolution sequence data, the evolution direction and evolution speed of the abnormal diagnosis region are calculated, and an evolution trend vector is generated.

[0088] Specifically, this is achieved by calculating the displacement of adjacent time phases and their intervals with real-world time points. The ratio of (physical time between two phases) represents the evolution rate;

[0089] ;in, , ;

[0090] ;in, It represents the direction of evolution.

[0091] In response to a user query, the system retrieves the coordinate data records of the fault source that conform to a specific spatial evolution pattern from the diagnostic area coordinate database and generates a visual analysis report.

[0092] 1. A topological network is constructed by utilizing the inherent vascular conduits (such as blood vessels, nerves, and ducts) within the tissue. Due to the relative constancy of the vascular system topology of biological tissues, the relative coordinates (distance and azimuth) of the lesion area relative to the nearest branch node are calculated, effectively offsetting the positioning errors caused by slice rotation, scaling, and tissue deformation, thus ensuring the spatial alignment of the lesion area at different time points.

[0093] 2. By associating abnormal diagnostic areas with the vascular topology network, this invention transforms isolated marker points into anatomically significant fault source coordinates. This recording method, using blood vessels as coordinate axes, can clearly reveal the spatial logic of tumor cell infiltration along the vascular system, providing researchers with a standardized data index for reproducing tumor spread pathways and analyzing microenvironment relationships.

[0094] 3. Based on the evolution trend vector generated by the topological network, the system can automatically calculate the evolution speed and direction of lesions. When responding to user queries, the system can generate a structured and visual analysis report, transforming complex cell morphological evolution into intuitive vector dynamics, allowing reviewers to clearly grasp the evolution stage of the lesion at a glance, significantly reducing the cognitive load of manual verification.

[0095] 4. Through anomaly region identification and connected component analysis, traditional large-size whole-slice image data can be transformed into structured, lightweight coordinate records and feature vectors. This data transformation significantly compresses storage space and establishes an indexing system centered on anatomical topological relationships. Therefore, the database can support efficient and accurate querying and aggregation of complex spatial evolution patterns (such as "evolution along blood vessels towards distal ends").

[0096] In the description of this specification, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0097] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0098] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for marking and archiving diagnostic areas on pathological slides, characterized in that, The method is executed by a processor and includes the following steps: Acquire multi-temporal pathological slide image data of the target tissue region, wherein the multi-temporal pathological slide image data includes slide images of the target tissue region collected at different time points; Based on the multi-temporal pathological slice image data, a cable conduit topology network is constructed within the target tissue area using an image registration algorithm and a topology refinement algorithm. The cable conduit topology network includes a cable conduit skeleton characterized by a centerline and multiple branch nodes located on the cable conduit skeleton. For the pathological slide image data of each time phase, abnormal diagnostic areas in the pathological slide image data are identified by an image recognition algorithm; For each time phase, the identified abnormal diagnostic area is used to calculate the geometric topological relationship data between its spatial location and the nearest neighboring branch node in the cable duct topology network. The geometric topological relationship data includes distance value and relative azimuth angle. Based on the geometric topology data, fault source coordinate data corresponding one-to-one with the abnormal diagnosis area is generated. The fault source coordinate data uses the branch nodes in the cable duct topology network as spatial reference benchmarks and is stored in the diagnosis area coordinate database.

2. The method for marking and archiving diagnostic areas of pathological slides according to claim 1, characterized in that, The steps for constructing the cable conduit topology network are as follows: The multi-temporal pathological slice image data is segmented to extract cable and pipe structure images in the target tissue region; The cable duct structure image is processed using a topology thinning algorithm to obtain a cable duct skeleton with a single pixel width; Identify pixels in the cable conduit skeleton with a connection count greater than 2 and determine them as branch nodes.

3. The method for marking and archiving diagnostic areas of pathological slides according to claim 1, characterized in that, The method for identifying the abnormal diagnostic region is as follows: Using an image segmentation algorithm based on gradient or intensity threshold, connected regions that are significantly different from the surrounding tissue are segmented from the pathological slide image data, and these connected regions are marked as the abnormal diagnostic regions.

4. The method for marking and archiving diagnostic areas of pathological slides according to claim 1, characterized in that, The method for calculating the relative azimuth angle is as follows: Taking the nearest neighboring branch node as the origin and the tangent direction of the skeleton segment where the node is located as the reference axis, calculate the angle between the direction vector of the centroid of the abnormal diagnosis region relative to the origin and the reference axis, and determine the angle as the relative azimuth angle.

5. The method for marking and archiving diagnostic areas of pathological slides according to claim 1, characterized in that, The distance value is calculated as follows: Calculate the Euclidean geometric distance from the centroid of the abnormal diagnostic region to its nearest branch node, and determine the Euclidean geometric distance as the distance value.

6. The method for marking and archiving diagnostic areas of pathological slides according to claim 1, characterized in that, Storing the fault source coordinate data into the diagnostic area coordinate database includes: The fault source coordinate data, the corresponding time phase information, and the corresponding target organization area identifier are associated and stored to form a structured database record.

7. The method for marking and archiving diagnostic areas of pathological slides according to claim 3, characterized in that, The specific steps for obtaining the connected region are as follows: The pathological slide image data is converted into first grayscale image data; Apply a gradient-based edge detection algorithm or a pixel intensity threshold-based binarization algorithm to the first grayscale image data to generate first binarized image data. The first pixel value in the first binarized image data represents a candidate region that differs from the surrounding tissue, and the second pixel value represents the background region. Perform connectivity analysis on the first binarized image data to obtain at least one candidate connected region data; Based on a preset minimum area threshold, candidate connected regions with an area greater than the minimum area threshold are selected from the at least one candidate connected region data, and the selected candidate connected region data is marked as the abnormal diagnosis region data.

8. The method for marking and archiving diagnostic areas of pathological slides according to claim 1, characterized in that, The method further includes: The fault source coordinate data generated at different times for the same target organization region are correlated in time to form the evolution sequence data of the abnormal diagnosis region; Based on the changes in the coordinate data of the fault source in adjacent time phases of the evolution sequence data, the evolution direction and evolution speed of the abnormal diagnosis region are calculated, and an evolution trend vector is generated. In response to a user query, the system retrieves the coordinate data records of the fault source that conform to a specific spatial evolution pattern from the diagnostic area coordinate database and generates a visual analysis report.

Images