Pathological index determination method, device, equipment and medium

Abstract

This application discloses a method, apparatus, device, and storage medium for determining pathological indicators. The method includes: acquiring a bone marrow pathological image to be processed; preprocessing the bone marrow pathological image to obtain a set of bone marrow sample images; processing the set of bone marrow sample images to obtain a bone marrow tissue region image; processing the bone marrow tissue region image to determine the bone marrow adipose tissue region and trabecular bone tissue region, removing the bone marrow adipose tissue region and trabecular bone tissue region to obtain a processed bone marrow tissue image; inputting the processed bone marrow tissue image into a trained image segmentation model to obtain the bone marrow fibrous region; and statistically analyzing the degree of fibrosis based on the bone marrow tissue region image and the bone marrow fibrous region to determine pathological indicators. This method can extract the layout details of the pathological image, thereby enabling a more comprehensive and fine-grained statistical analysis of the degree of fibrosis and improving the accuracy of pathological indicator determination.

Description

Technical Field

[0001] This invention generally relates to the field of medical technology, and specifically to a method, apparatus, equipment and medium for determining pathological indicators. Background Technology

[0002] With the continuous development of medical technology, analysis and screening based on pathological images are important means of disease diagnosis and treatment in modern medicine. For example, the analysis of bone marrow slice images generally focuses on the distribution and morphology of bone marrow tissue slice images. By observing the diseased tissue, pathological indicators of myelofibrosis can be determined, thus providing auxiliary means for drug trials, postoperative analysis, medical planning, evaluation of treatment effects, and other stages.

[0003] Currently, the relevant technology uses manual observation based on bone marrow pathological images to determine myelofibrosis pathological indicators. However, this method relies solely on human knowledge and experience in reading images, lacking objectivity and resulting in low accuracy in determining myelofibrosis pathological indicators. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method, device, equipment and medium for determining pathological indicators.

[0005] In a first aspect, the present invention provides a method for determining pathological indicators, the method comprising:

[0006] Acquire bone marrow pathology images to be processed;

[0007] The bone marrow pathology images to be processed are preprocessed to obtain a set of bone marrow sample images;

[0008] The bone marrow sample image set is processed to obtain bone marrow tissue region images;

[0009] Image processing is performed on the bone marrow tissue region image to determine the bone marrow adipose tissue region and the trabecular bone tissue region. The bone marrow adipose tissue region and the trabecular bone tissue region are then removed to obtain the processed bone marrow tissue image.

[0010] The processed bone marrow tissue image is input into a trained image segmentation model to obtain a binary image. The contour of the binary image is calculated to obtain the bone marrow fibrous region.

[0011] Based on the bone marrow tissue region images and the bone marrow fibrosis region, the degree of fibrosis was statistically analyzed to determine pathological indicators.

[0012] Secondly, embodiments of this application provide a pathological index determination device, the device comprising:

[0013] The acquisition module is used to acquire bone marrow pathology images to be processed;

[0014] The preprocessing module is used to preprocess the bone marrow pathology images to be processed to obtain a set of bone marrow sample images;

[0015] The first processing module is used to perform image processing on the bone marrow sample image set to obtain bone marrow tissue region images;

[0016] The second processing module is used to perform image processing on the bone marrow tissue region image, determine the bone marrow adipose tissue region and the trabecular bone tissue region, remove the bone marrow adipose tissue region and the trabecular bone tissue region, and obtain the processed bone marrow tissue image.

[0017] The fiber region determination module is used to input the processed bone marrow tissue image into a trained image segmentation model to obtain a binary image, calculate the contour of the binary image, and obtain the bone marrow fiber region.

[0018] The statistical analysis module is used to statistically analyze the degree of fibrosis and determine pathological indicators based on the bone marrow tissue region image and the bone marrow fibrosis region.

[0019] Thirdly, embodiments of this application provide an apparatus including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the pathological indicator determination method as described in the first aspect above.

[0020] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon for implementing the pathological indicator determination method of the first aspect above.

[0021] The pathological index determination method, apparatus, equipment, and medium provided in this application embodiment acquire bone marrow pathological images to be processed, preprocess these images to obtain a set of bone marrow sample images, then process these images to obtain bone marrow tissue region images, further process these images to determine the bone marrow adipose tissue region and trabecular bone tissue region, remove the adipose tissue and trabecular bone tissue regions to obtain a processed bone marrow tissue image, input the processed bone marrow tissue image into a trained image segmentation model to obtain a binary image, calculate the contour of the binary image to obtain the bone marrow fibrous region, and finally, statistically analyze the degree of fibrosis based on the bone marrow tissue region image and the bone marrow fibrous region to determine pathological indexes. This technical solution, on the one hand, by preprocessing the bone marrow pathological images to be processed, can accurately determine the set of bone marrow sample images and the bone marrow tissue region images, providing data guidance for subsequent determination of the bone marrow adipose tissue region and trabecular bone tissue region. On the other hand, by processing the bone marrow tissue images through a trained image segmentation model, the layout details of the pathological images can be extracted, thereby accurately and quickly obtaining binary images. This allows for precise identification of the bone marrow fibrosis region, enabling more comprehensive and fine-grained statistical analysis of the degree of fibrosis based on the bone marrow tissue region images and the bone marrow fibrosis region. This provides doctors with accurate quantitative results, reduces the instability of pathological indicators caused by subjective human factors, assists doctors in making efficient and accurate judgments, and improves the accuracy of pathological indicator determination. Attached Figure Description

[0022] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0023] Figure 1 A schematic diagram of the implementation environment for the pathological index determination method provided in the embodiments of this application;

[0024] Figure 2 A flowchart illustrating the method for determining pathological indicators provided in the embodiments of this application;

[0025] Figure 3 This is a schematic diagram illustrating the structure of an argyrophilic section of a bone marrow sample provided in an embodiment of this application.

[0026] Figure 4 This is a schematic diagram of the structure for preprocessing bone marrow pathology images to be processed, provided in an embodiment of this application.

[0027] Figure 5 A schematic flowchart illustrating the method for determining argyrophilic sections of bone marrow tissue provided in this application embodiment;

[0028] Figure 6This is a schematic diagram of the structure of the bone marrow tissue region provided in an embodiment of this application;

[0029] Figure 7 A schematic diagram of the structure of the bone marrow adipose tissue region and the bone marrow trabecular bone tissue region provided in the embodiments of this application;

[0030] Figure 8 A schematic diagram of the structure of the single fat droplet region and the fat aggregation region provided in the embodiments of this application;

[0031] Figure 9 This is a schematic diagram illustrating the acquisition of adipose-free argyrophilic sections according to an embodiment of this application;

[0032] Figure 10 A schematic diagram illustrating the acquisition of argyrophilic sections of trabecular bone tissue provided in an embodiment of this application;

[0033] Figure 11 This is a schematic diagram of the structure of the image segmentation model provided in the embodiments of this application;

[0034] Figure 12 This is a schematic diagram of the structure of the bone marrow fibrous region provided in an embodiment of this application;

[0035] Figure 13 A flowchart illustrating the method for constructing an image segmentation model provided in an embodiment of this application;

[0036] Figure 14 A schematic diagram illustrating the determination of myelofibrosis results provided in an embodiment of this application;

[0037] Figure 15 A flowchart illustrating the method for calculating the degree of fibrosis in bone marrow tissue segments provided in this application embodiment;

[0038] Figure 16 A schematic diagram illustrating the calculation of parameters for the degree of fibrosis in bone marrow tissue segments, provided in an embodiment of this application;

[0039] Figure 17 A schematic diagram of the structure of the segmented bone marrow tissue region image provided in the embodiments of this application.

[0040] Figure 18 This is a schematic diagram of the structure of the method for determining pathological indicators provided in the embodiments of this application;

[0041] Figure 19 This is a schematic diagram of the pathological index determination device provided in the embodiments of this application;

[0042] Figure 20 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0043] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0044] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0045] Understandably, with the rapid development of artificial intelligence (AI) technology, its integration with medicine has been applied in multiple medical fields, including medical imaging, assisted diagnosis, drug development, health management, disease prediction, hospital management, virtual assistants, medical robots, and medical research. In clinical practice, analyzing the distribution and morphology of pathological images is of significant value in disease diagnosis. For example, during the examination of myelofibrosis, the distribution and morphology of bone marrow pathological images can be analyzed to determine pathological indicators.

[0046] It is important to note that myelofibrosis, or myelofibrosis for short, is a myeloproliferative disorder caused by collagen proliferation in the bone marrow hematopoietic tissue. This fibrous tissue severely affects hematopoietic function, leading to the replacement of normal hematopoietic tissue by proliferating fibrous tissue for known or unknown reasons. This results in abnormal bone marrow hematopoietic function and a series of clinical symptoms. The disease is mainly characterized by varying degrees of cytopenia or cytosis, with erythroblasts and myelocytes visible in peripheral blood, along with numerous teardrop-shaped red blood cells. Extramedullary hematopoiesis and bone marrow tapping are observed, often accompanied by varying degrees of hepatosplenomegaly. Bone marrow histological examination is an invasive diagnostic procedure. This examination requires a puncture at the bone marrow site, followed by tissue extraction and smear analysis. Bone marrow histological examination is of significant value in diagnosing myelofibrosis. If a patient has myelofibrosis, the problem can be quickly identified through visualization of the bone marrow section, and it can also be determined whether there is proliferation or atrophy of local fibrous tissue.

[0047] Currently, one approach in this field involves manually observing bone marrow pathology images with the naked eye to examine the distribution or morphology of myelofibrosis and determine pathological indicators. However, this method relies solely on human knowledge and experience in interpreting images, which is not only susceptible to the subjective limitations of physicians' experience, but also, with the increase in clinical sample sizes, routine manual image interpretation can no longer meet the clinical needs of hospital pathology departments, resulting in low accuracy in determining myelofibrosis pathological indicators.

[0048] To address the aforementioned deficiencies, this application provides a method, apparatus, device, and medium for determining pathological indicators. Compared with existing technologies, this solution, on the one hand, preprocesses the bone marrow pathological images to be processed, enabling precise identification of the bone marrow sample image set and bone marrow tissue region images, providing data guidance for subsequent identification of bone marrow adipose tissue regions and trabecular bone tissue regions. On the other hand, by processing the processed bone marrow tissue images using a trained image segmentation model, the layout details of the pathological images can be extracted, thereby accurately and quickly obtaining binary images. This allows for precise identification of bone marrow fibrosis regions, enabling more comprehensive and fine-grained statistical analysis of the degree of fibrosis based on bone marrow tissue region images and bone marrow fibrosis regions. This provides doctors with accurate quantitative results, reduces the instability of pathological indicators due to subjective human factors, assists doctors in making efficient and accurate judgments, and improves the accuracy of pathological indicator determination.

[0049] Figure 1 This is an implementation environment architecture diagram of a method for determining pathological indicators provided in an embodiment of this application. For example... Figure 1 As shown, the implementation environment architecture includes: terminal 100 and server 200.

[0050] In the field of image processing, the image processing of the bone marrow pathology image to be processed can be performed on the terminal 100 or on the server 200. For example, the bone marrow pathology image to be processed can be acquired through the terminal 100, and image processing can be performed locally on the terminal 100 to obtain the pathological indicators of the bone marrow pathology image; alternatively, the bone marrow pathology image to be processed can be sent to the server 200, so that the server 200 can acquire the bone marrow pathology image to be processed, perform image processing based on the bone marrow pathology image to obtain the pathological indicators of the bone marrow pathology image to be processed, and then send the pathological indicators of the bone marrow pathology image to the terminal 100 to realize the process of determining the pathological indicators of the bone marrow pathology image to be processed.

[0051] In addition, the terminal 100 may run an operating system, which may include, but is not limited to, Android, iOS, Linux, Unix, Windows, etc. It may also include a user interface (UI) layer, which can provide the display of the bone marrow pathology images to be processed and the pathological indicators of the bone marrow pathology images to be processed. In addition, the bone marrow pathology images to be processed required for image processing can be sent to the server 200 based on the application programming interface (API).

[0052] Optionally, terminal 100 can be a terminal device in various AI application scenarios. For example, terminal 100 can be a laptop, tablet, desktop computer, in-vehicle terminal, intelligent voice interaction device, smart home appliance, mobile device, aircraft, etc. Mobile devices can be various types of terminals such as smartphones, portable music players, personal digital assistants, dedicated messaging devices, portable gaming devices, etc. This application embodiment does not specifically limit this.

[0053] Server 200 can be a single server, a server cluster or distributed system consisting of several servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0054] Terminal 100 and server 200 establish a communication connection via a wired or wireless network. Optionally, the aforementioned wireless or wired network uses standard communication technologies and / or protocols. The network is typically the Internet, but can also be any network, including but not limited to a Local Area Network (LAN), Metropolitan Area Network (MAN), Wide Area Network (WAN), mobile, wired or wireless network, private network, or any combination of virtual private network.

[0055] The aforementioned deep learning-based AI application system, in providing AI application services, can extract features and segment bone marrow pathology images to be processed using an image segmentation model to obtain bone marrow fibrosis regions. It can then determine pathological indicators based on the degree of fibrosis in these regions to provide AI application services. The image segmentation model can be located in server 200, trained and applied by the server; alternatively, the image segmentation model can be located in terminal 100, trained and updated by server 200.

[0056] For ease of understanding and explanation, the following will use... Figures 2 to 20 This application provides a detailed description of the methods, apparatus, equipment, and storage media for determining pathological indicators provided in its embodiments.

[0057] Figure 2 The diagram shown is a schematic flowchart of a pathological index determination method according to an embodiment of this application. This method can be executed by a pathological index determination device. Figure 2 As shown, the method includes:

[0058] S101. Obtain the bone marrow pathology image to be processed.

[0059] The aforementioned bone marrow pathology images to be processed refer to bone marrow pathology images for which pathological indicators need to be determined. These images can be original bone marrow slice images, obtained by scanning silver-stained clinical biopsy bone marrow samples using a scanner. The number of such bone marrow pathology images to be processed can be one or more.

[0060] For clinical biopsy bone marrow samples, those meeting the requirements for bone marrow sample preparation are generally required to be 1.5-2.5 cm in length; specimens smaller than 1.5 cm are considered unacceptable. The longer the biopsy specimen, the larger the observation range, the more representative it is, and the higher the reliability of pathological marker determination. The processed bone marrow pathological images have corresponding layers; generally, the higher the layer, the larger the image size; the lower the layer, the smaller the image size.

[0061] The bone marrow pathology image to be processed may include areas of adipose tissue, areas of trabecular bone tissue, or areas of fibrous tissue.

[0062] In one possible implementation, the acquisition of the aforementioned bone marrow pathology images may include: obtaining microscopic images through an image scanner, digital camera, etc., and then converting the microscopic images into electrical signals and digitizing them. In practice, biological methods such as the preparation of microscopic samples (smears, sections) are also included before image acquisition. In another possible implementation, the images can be directly imported from external devices, or obtained from a pre-set database or blockchain.

[0063] S102. Preprocess the bone marrow pathology images to be processed to obtain a set of bone marrow sample images.

[0064] Due to the complexity of the bone marrow pathology images to be processed, as well as external interference during the sampling process, the accurate segmentation of bone marrow pathology images can be affected. Therefore, it is necessary to perform necessary preprocessing before segmentation to smooth and eliminate noise, enhance the distinction between the target and the background, and eliminate changes in illumination field caused by uneven light source.

[0065] The aforementioned bone marrow sample image set refers to bone marrow sample images that meet the requirements of standard bone marrow sample slides. After obtaining the bone marrow pathology image to be processed, which may be an image in MRXS format, the image can be preprocessed to remove irrelevant areas of the bone marrow sample, resulting in a bone marrow sample image set that meets the bone marrow slide preparation standards.

[0066] Please see Figure 3As shown, taking the bone marrow pathology image to be processed as the original bone marrow argyrophilic section and the set of bone marrow sample images as the bone marrow sample argyrophilic section as an example, in the preprocessing of the original bone marrow argyrophilic section, irrelevant regions of bone marrow tissue can be filtered first to obtain filtered bone marrow section images. It is then determined whether the filtered bone marrow section images meet the bone marrow preparation standards. If they meet the standards, the contour coordinates of the compliant bone marrow section images are mapped to the corresponding level of the original bone marrow argyrophilic section, thus obtaining the bone marrow sample argyrophilic section. If they do not meet the standards, a bone marrow preparation request is generated based on the compliant bone marrow argyrophilic section and sent to the client, resulting in a set of bone marrow sample argyrophilic sections that meet the standards. (See also...) Figure 4 As shown, where, Figure 4 The image on the left is the original argyrophilic section of bone marrow, and the image on the right is a collection of argyrophilic sections of bone marrow samples that meet the bone marrow preparation standards after preprocessing.

[0067] In this embodiment, during the preprocessing of the bone marrow pathology image to be processed, the MRXS format image can be converted to RGBA format. Then, the pixel values ​​of the A channel in the RGBA format image are obtained. Based on the A channel pixel values ​​and a preset A channel threshold, the bone marrow pathology image is segmented to obtain multiple bone marrow samples. It is then determined whether each bone marrow sample meets the bone marrow slide preparation standards. Each bone marrow sample can be filtered according to a preset size pixel threshold to obtain a set of bone marrow sample images. The preset A channel threshold is a custom-set slice transparency threshold based on actual needs, and the size pixel threshold is the pixel value corresponding to the size that meets the bone marrow slide preparation standards, set according to actual needs.

[0068] In the process of segmenting bone marrow pathology images based on the pixel values ​​of the A channel and a preset A channel threshold, the pixel values ​​of the A channel can be compared with the preset A channel threshold. Regions in the RGBA format bone marrow image where the pixel values ​​of the A channel are greater than the preset A channel threshold are identified as multiple bone marrow samples. Then, for each of these multiple bone marrow samples, the pixel values ​​are obtained and compared with a size pixel threshold. When a pixel value is greater than the size pixel threshold, the bone marrow sample with a value greater than the size pixel threshold is retained, indicating that it meets the bone marrow slide preparation standard. The contour coordinates of this compliant bone marrow sample are mapped to the corresponding layer of the original argyrophilic bone marrow section, thus obtaining a set of bone marrow sample images. When a pixel value is not greater than the size pixel threshold, the bone marrow sample is determined to not meet the bone marrow slide preparation standard. A bone marrow slide preparation request is generated based on the bone marrow samples with pixel values ​​not greater than the size pixel threshold, and the request is resent to the client. The client then responds to the request by correcting the bone marrow samples with pixel values ​​not greater than the size pixel threshold, thus obtaining bone marrow samples that meet the slide preparation standard.

[0069] It should be noted that the RGBA format represents a color space consisting of Red, Green, Blue, and Alpha. Although it is sometimes described as a color space, it is actually just an additional information added to the RGB model. It can belong to any RGB color space. The Alpha channel (A channel) is generally used as an opacity parameter. If the Alpha channel value of a pixel is 0%, it is completely transparent (i.e., invisible), while a value of 100% means a completely opaque pixel (in traditional digital images). Values ​​between 0% and 100% allow pixels to show through the background. Therefore, in this embodiment, a preset A channel threshold can be compared with the pixel values ​​of the A channel to segment the bone marrow pathology image to be processed, thereby accurately obtaining multiple bone marrow samples.

[0070] In this embodiment, by determining whether the bone marrow pathology image to be processed meets the bone marrow preparation standards, a set of bone marrow sample images that meet the bone marrow preparation standards can be accurately obtained, thereby enabling targeted subsequent image processing and providing data guidance information for subsequent image processing.

[0071] S103. Perform image processing on the bone marrow sample image set to obtain bone marrow tissue region images.

[0072] The aforementioned bone marrow tissue region image refers to a sample image including the bone marrow tissue region. After obtaining the set of bone marrow sample images, image processing can be performed on each bone marrow sample image in the set to obtain the bone marrow tissue region image. Specifically, for each bone marrow sample image in the set, the bone marrow sample image is first converted into a BGR format bone marrow tissue image. Then, channel separation processing is performed on the bone marrow tissue image to obtain B-channel, R-channel, and G-channel images. Based on the B-channel, R-channel, and G-channel images, they are converted into binarized images. Finally, based on the B-channel, R-channel, and binarized images, edge optimization and contour extraction processing are performed on the bone marrow tissue image to obtain the bone marrow tissue region image.

[0073] Specifically, the bone marrow sample image mentioned above is in RGBA format. This RGBA format bone marrow sample image can be converted to a BGR format bone marrow tissue image. Then, the BGR format bone marrow tissue image can be processed by channel separation, for example, by calling the split function in OpenCV to separate the channels and obtain B channel image, R channel image and G channel image. Then, the pixel values ​​of the B channel image, R channel image and G channel image can be obtained. Then, the pixel values ​​of the B channel image, R channel image and G channel image are compared with a preset pixel threshold and binarized to obtain a binarized image.

[0074] It should be noted that a binarized image means that each pixel in the image has only two possible values ​​or grayscale levels. That is, the grayscale value of any pixel in the image is either 0 or 255, representing black and white, respectively.

[0075] It is understandable that the acquired bone marrow sample images may be affected by noise such as jagged edges due to changes in illumination field caused by uneven light source. Therefore, after obtaining the B-channel image, R-channel image, G-channel image and binarized image, pixel-level operations can be used to optimize the edges of the bone marrow tissue image to reduce the jaggedness of the bone marrow tissue image, thereby achieving image smoothing.

[0076] Optionally, during the edge optimization of bone marrow tissue images, the Laplacian operator can be used to enhance the B-channel, R-channel, and G-channel images separately. This can be achieved by performing multiple pixel-level dilation and erosion cross operations on the binarized image, such as three pixel-level dilation and erosion cross operations.

[0077] After edge optimization processing of the bone marrow tissue image, the contour information of the bone marrow tissue region image is calculated and obtained. The B-channel image, R-channel image and binarized image are overlaid and thresholded for filtering analysis to obtain the bone marrow processed image. Finally, the contour coordinates of the bone marrow processed image are mapped to the corresponding layer of the argyrophilic slice of the bone marrow sample to obtain the bone marrow tissue region image.

[0078] The above-mentioned method of obtaining a bone marrow-processed image by overlaying and thresholding the B-channel image, R-channel image, and binarized image can be achieved by comparing the pixel values ​​of the B-channel image with a first threshold, the pixel values ​​of the R-channel image with a second threshold, and the pixel values ​​of the binarized image with a third threshold. Regions where the pixel values ​​of the B-channel image are greater than the first threshold, the pixel values ​​of the R-channel image are greater than the second threshold, and the pixel values ​​of the binarized image are greater than the third threshold are set to 1, thus identifying the image as a bone marrow-processed image; otherwise, they are set to 0. This can be expressed by the following formula:

[0079]

[0080] Where R represents the pixel values ​​of the R-channel image, B represents the pixel values ​​of the B-channel image, and img bin Mask represents the pixel values ​​of a binary image. capillary_tuft Image processing for bone marrow.

[0081] Contour coordinate mapping methods for bone marrow processed images refer to the smooth transition of the image contour from a low-level to a high-level contour, preventing jagged edges. Linear interpolation algorithms are generally used for contour coordinate mapping, such as unilinear and bilinear interpolation. Unilinear interpolation can be expressed by the following formula:

[0082]

[0083] Where (x0, y0) and (x1, y1) are the coordinates of the data points of the contour on the bone marrow processed image, and (x, y) are the coordinates of the data points of the contour on the bone marrow tissue region image.

[0084] Bilinear interpolation can be expressed by the following formula:

[0085]

[0086] Where (x, y) are the coordinates of the data points on the contour of the bone marrow tissue region image.

[0087] Please see Figure 5As shown, when the bone marrow sample image set is a set of argyrophilic slices of bone marrow samples, and the bone marrow tissue region image is an argyrophilic slice of bone marrow tissue, for the argyrophilic slices of bone marrow samples in the set of argyrophilic slices of bone marrow samples, firstly, the specified level of the argyrophilic slices of bone marrow samples is read, and the argyrophilic slices of bone marrow samples are converted into bone marrow tissue images in BGR format. Secondly, the bone marrow tissue images are processed by channel separation to obtain B-channel images, R-channel images, and G-channel images. At the same time, the B-channel images, R-channel images, and G-channel images are converted into binarized images. Then, pixel-level operations are used to perform edge calculation processing or bone marrow tissue contour optimization processing at different levels on the bone marrow tissue images. Finally, the contour information of the optimized bone marrow tissue image is calculated. The contour information of this bone marrow tissue image can be found in [reference needed]. Figure 6 As shown, the contour coordinates of the bone marrow tissue image are then obtained from the contour information, and the contour coordinates are mapped to the corresponding layer of the bone marrow argyrophilic section to obtain the bone marrow tissue argyrophilic section.

[0088] It should be noted that during the process of performing multiple pixel-level dilation and erosion operations, dilation can be understood as dealing with image defects, and erosion can be understood as dealing with image artifacts. The dilation or erosion operation involves convolving the image (or a portion of the image) with a structuring element (convolution kernel) to calculate the maximum value of the pixels in the area covered by the convolution kernel, and then assigning this maximum value to the pixel specified by the reference point.

[0089] The core of dilation and erosion operations is the structuring element. Generally, the structuring element is composed of a matrix with elements of 1 or 0. The region with a structuring element of 1 defines the neighborhood of the image, and the pixels within the neighborhood must be considered when performing morphological operations such as dilation and erosion.

[0090] Erosion refers to removing some burrs and details from an image. It is generally used to eliminate noise and segment individual image elements. Essentially, it is also a kind of spatial filtering. A mask is set, and the center of the mask passes through each pixel point in turn. The value of the current pixel point (i.e., the position corresponding to the center of the mask) is set as the minimum value of the pixels in the mask coverage area.

[0091] The erosion operation involves convolving an image (or a region of an image) with a structuring element (a convolution kernel). The kernel can be of any shape and size and has a separately defined reference point. In most cases, the kernel is a small square or disk with a reference point in the center; it can be considered a template or mask. The pixel at the center of the mask is checked against the surrounding pixels (i.e., whether it is white, or its pixel value is 255). If they match, the pixel is retained; otherwise, it is set to black (i.e., its pixel value is set to 0).

[0092] As an optional implementation, in the process of performing three dilation and erosion operations on a binary image, the dilation operation can be performed first, followed by the erosion operation; then, the dilation operation is performed a second time, followed by the erosion operation; and finally, the dilation operation is performed a third time, followed by the erosion operation. Specifically, the dilation operation uses a structuring element (typically 3×3 in size) to scan every pixel in the image. Each pixel in the structuring element is then ANDed with the pixels it covers. If both values ​​are 0, the pixel value is set to 0; otherwise, it is set to 1. Similarly, the erosion operation uses a structuring element (typically 3×3 in size) to scan every pixel in the image. Each pixel in the structuring element is then ANDed with the pixels it covers. If both values ​​are 1, the pixel value is set to 1; otherwise, it is set to 0.

[0093] In this embodiment, by performing edge calculation and contour optimization of bone marrow tissue, the influence of image noise can be suppressed, making the determined bone marrow tissue region image smoother and more accurate in determining the bone marrow adipose tissue region and trabecular bone tissue region.

[0094] S104. Perform image processing on the bone marrow tissue region image to determine the bone marrow adipose tissue region and the trabecular bone tissue region, remove the bone marrow adipose tissue region and the trabecular bone tissue region to obtain the processed bone marrow tissue image.

[0095] The aforementioned bone marrow adipose tissue region refers to the area in a bone marrow tissue image that includes adipose tissue. The aforementioned trabecular bone tissue region refers to the area in a bone marrow tissue image that includes trabecular bone tissue. The aforementioned processed bone marrow tissue image refers to an image of a bone marrow tissue region after removing the bone marrow adipose tissue region and the trabecular bone tissue region.

[0096] One possible implementation involves first using image processing algorithms to process the bone marrow tissue region image to identify the bone marrow adipose tissue region and the trabecular bone tissue region. Then, the bone marrow adipose tissue region and the trabecular bone tissue region are removed from the bone marrow tissue region image, resulting in a processed bone marrow tissue region image. (See also...) Figure 7 As shown, Figure 7 The darker areas in the medium color represent the bone marrow adipose tissue region and trabecular bone tissue region located in the bone marrow tissue region image.

[0097] Another possible implementation is to first perform image processing on the bone marrow tissue region image to determine the bone marrow adipose tissue region, then remove the bone marrow adipose tissue region to obtain a bone marrow tissue region image with the bone marrow adipose tissue region removed, and then further process the image to determine the trabecular bone tissue region, remove the trabecular bone tissue region, thereby obtaining a processed bone marrow tissue region image.

[0098] Specifically, after identifying the bone marrow tissue region image, the image undergoes format conversion, block processing, and image segmentation to accurately locate single fat droplets and multiple fat aggregation regions within the bone marrow tissue region image, thus obtaining the bone marrow adipose tissue region. This bone marrow tissue region image can be in BGR format.

[0099] As an optional implementation, the bone marrow tissue region image can first be converted into a LAB format image. This LAB format image can then be divided into blocks of a fixed size, resulting in multiple block images. For each block image, the pixel value in the L color channel is determined. Based on the L color channel pixel value and a preset pixel threshold, a fat-like region is identified. Based on the morphological characteristics of this fat-like region, single fat droplet regions and multiple fat aggregation regions are determined. Finally, based on these single fat droplet regions and multiple fat aggregation regions, the bone marrow adipose tissue region is determined. The aforementioned fat-like regions include single fat droplet regions, multiple fat aggregation regions, and other regions.

[0100] Optionally, in the process of screening for adipose regions and identifying adipose-like regions from bone marrow tissue images, the following formula can be used:

[0101]

[0102] Where L is the pixel value of the L color channel, and Mask fat This is a fat-like region.

[0103] It should be noted that LAB is a less commonly used color space, a color system based on physiological characteristics. In the LAB color space, the L component represents pixel brightness, with a value range of [0, 100], representing pure black to pure white; A represents the range from red to green, with a value range of [127, -128]; and B represents the range from yellow to blue, with a value range of [127, -128]. The BGR channel mentioned above refers to a channel composed of the red channel (R), green channel (G), and blue channel (B).

[0104] The process involves first converting BGR format bone marrow tissue images to XYZ format, and then converting the XYZ format images to LAB format. BGR format values ​​range from 0 to 255; converting BGR to XYZ means normalizing 0-255 to 0-1. The LAB format bone marrow tissue images are then segmented into multiple blocks of a fixed size. For each block, the pixel value in the L color channel is determined, and this value is compared to a preset pixel threshold range. Regions where the L color channel pixel value falls within this threshold range are identified as fat-like regions. Morphological features of these fat-like regions are then acquired, such as the area ratio within the circumcircle of the fat-like region. After identifying the fat-like regions, single fat droplet regions and multiple fat aggregation regions are determined based on their morphological features to avoid mis-screening of surrounding fibers. This can be achieved using the area ratio within the circumcircle of the fat-like region. bone Threshold filtering was performed to obtain single-droplet regions and multiple-fat-aggregate regions, which were then identified as bone marrow adipose tissue regions. The single-droplet regions and multiple-fat-aggregate regions can be represented by the following formula:

[0105]

[0106] Please see Figure 8 As shown, Figure 8 This is a schematic diagram of the structure of fat. Figure 8 The left side shows a structural diagram of a single fat droplet region, while the right side shows a structural diagram of a multi-fat aggregation region.

[0107] Please see Figure 9 As shown, taking a bone marrow tissue argyrophilic section as an example, after obtaining the argyrophilic section, it can be first divided into blocks of fixed size to obtain multiple block images. Then, bone marrow adipose tissue calculations are performed to determine the bone marrow adipose tissue region, which is then removed to obtain a bone marrow section without adipose tissue. Furthermore, after determining the bone marrow adipose tissue region, its contour coordinates can be mapped to the corresponding layer to obtain the corresponding bone marrow adipose tissue region image.

[0108] As another possible approach, image processing can be performed on the bone marrow tissue region image to determine the trabecular bone tissue region. This can be achieved by performing image enhancement processing on the bone marrow tissue region image to obtain an enhanced image, converting the enhanced image into a binary image of the bone marrow mask, then calculating the pixel values ​​of all contours in the binary image of the bone marrow mask, and filtering the binary image of the bone marrow mask according to a first size threshold, a second size threshold, and the pixel values ​​to obtain complete trabecular bone regions and trabecular bone fragment regions, wherein the first size threshold is greater than the second size threshold. Finally, based on the complete trabecular bone regions and trabecular bone fragment regions, the trabecular bone tissue region is determined.

[0109] Please see Figure 10 As shown, taking the bone marrow tissue region image as an example of a argyrophilic section of bone marrow tissue, after obtaining the argyrophilic section of bone marrow tissue, the argyrophilic section of bone marrow tissue can first be stained to enhance the staining of the section. For example, the principal component colors purple and cyan in the argyrophilic section of bone marrow tissue can be enhanced to complete the significant characteristics of each component in the argyrophilic section of bone marrow tissue, thereby obtaining an enhanced image. Then, the bone marrow trabecular tissue is located to determine the bone marrow trabecular tissue region, and the bone marrow trabecular tissue region is removed to obtain a bone marrow section without bone trabecular tissue.

[0110] It should be noted that because the trabecular bone region has a relatively simple composition, its staining is relatively stable under standard conditions. The color of the bone marrow tissue region image can be adjusted according to a preset reference color to ensure the color of the trabecular bone region corresponds to the preset reference color, thus obtaining an enhanced image. The purpose of staining and enhancing the bone marrow tissue region image is to achieve bipolar differentiation of the corresponding colors of bone marrow fibers and trabecular bone. In determining the contour of the trabecular bone region based on the bone marrow tissue region image, the main process involves screening the block structures in the bone marrow tissue based on the size of the bone marrow to obtain trabecular bone regions that meet the standards. This can be expressed by the following formula:

[0111]

[0112] Among them, Ratio bone It represents the proportion of trabeculae in the entire bone marrow tissue region in the image.

[0113] Specifically, after obtaining the bone marrow tissue region image, the bone marrow tissue region image can first be stained and enhanced to strengthen the staining of the sections. Then, the bone marrow tissue region image is converted from BGR format to a bone marrow mask binary image. Then, the pixel values ​​of all contours in the bone marrow mask binary image are calculated, and the pixel values ​​of all contours are traversed. The first size threshold is compared with the pixel values. The regions in the bone marrow mask binary image with pixel values ​​greater than the first size threshold are identified as complete trabecular bone regions. The second size threshold is compared with the pixel values. The regions in the bone marrow mask binary image with pixel values ​​less than the second size threshold are identified as trabecular bone fragment regions. The complete trabecular bone regions and trabecular bone fragment regions are identified as trabecular bone tissue regions. The contour coordinates of the trabecular bone tissue are mapped to the corresponding levels. The trabecular bone tissue regions are removed, and then the argyrophilic bone marrow sections with trabecular bone tissue removed are obtained.

[0114] Optionally, during the filtering process of the binary image of the bone marrow mask based on the first size threshold, the second size threshold, and the pixel value, multiple size filters can be performed to determine the intact trabecular bone region and the trabecular bone fragment region. For example, this could involve three size filtering processes. By performing multiple size filtering processes, the accuracy of determining the intact trabecular bone region and the trabecular bone fragment region can be improved.

[0115] S105. Input the processed bone marrow tissue image into the trained image segmentation model to obtain a binary image. Calculate the contour of the binary image to obtain the bone marrow fibrous region.

[0116] It should be noted that the image segmentation model described above refers to a network structure model that learns the ability to determine binary images by training on sample data. The input of the image segmentation model is a processed bone marrow tissue image, and the output is the corresponding binary image. It also has the ability to perform image transformation on the processed bone marrow tissue image, making it a neural network model capable of determining the binary image. The image segmentation model can include a multi-layered network structure. Different layers of the network structure process the input data differently and transmit their output results to the next network layer, until the last network layer processes the data to obtain the binary image. This image segmentation model is responsible for establishing the relationship between the processed bone marrow tissue image and the binary image, and its model parameters are already in an optimal state.

[0117] This image segmentation model may include, but is not limited to, convolutional layers, normalization layers, and activation functions. These components may consist of a single layer or multiple layers. The convolutional layer is used to extract edge and texture features from the processed bone marrow tissue image. The normalization layer normalizes the image features obtained from the convolutional layer; for example, it can subtract the mean from the image features and divide by the variance to obtain a normal distribution with a mean of zero and a variance of one, thus preventing gradient explosion and gradient vanishing. The activation function can be a Sigmoid function, a Tanh function, or a ReLU function. By processing the normalized feature map through the activation function, the result can be mapped to the range of 0 to 1.

[0118] Since the processed bone marrow tissue image is too large to meet the input requirements of the image segmentation model, it is necessary to divide the processed bone marrow tissue image into blocks. This can be done by traversing the entire processed bone marrow tissue image and then dividing it into blocks according to a fixed size to obtain multiple block bone marrow tissue images.

[0119] Optionally, the image segmentation model mentioned above can be a UNet series model that has performed well in the field of artificial intelligence in medicine, such as UNet, UNet++, TransUNet, U2Net, UNeXt, and other models with relatively deep network layers. Among them, the lightweight medical image segmentation network UNeXt adopts a convolutional multilayer perceptron and a spatial attention mechanism. It uses a labeled multilayer perceptron module to optimize feature extraction, enabling the model to better focus on the macroscopic and microscopic information of the bone marrow fibrous region and improve the accuracy of target detection.

[0120] Understandably, taking UNeXt as an example, UNeXt is an encoder-decoder structure that includes a convolutional stage and a tokenized MLP stage. After obtaining multiple segmented bone marrow tissue images, each segmented bone marrow tissue image is input into the encoder of the image segmentation model to obtain a feature map. The feature map is then passed through the decoder of the image segmentation model to obtain a segmented binary image. Finally, the multiple segmented binary images are stitched together to obtain a binary image.

[0121] Specifically, please see Figure 11As shown, in the process of inputting each segment of bone marrow tissue image into the image segmentation model, each segment of bone marrow tissue image can first be passed through an encoder. The encoder structure consists of five modules: the first three modules are convolutional modules, and the last two modules are tokenized MLP modules. The decoder structure also consists of five modules: the first two modules are tokenized MLP modules, and the last three modules are convolutional modules. Each module in the encoder reduces the feature resolution by a factor of two, and each module in the decoder increases the resolution by two. Skip connections are also applied between the encoder and decoder.

[0122] The encoder's convolutional module has three modules, each consisting of a convolutional layer, a normalization layer, a pooling layer, and a ReLU activation layer. Specifically, the first convolutional module includes a Conv+BatchNorm+Pooling+ReLU layer, and the input segment of the bone marrow tissue image varies in size as follows: The second convolutional module consists of Conv+BatchNorm+Pooling+ReLU layers, and the size of its feature map varies as follows: The third convolutional module consists of Conv+BatchNorm+Pooling+ReLU layers, and the size of its feature map varies as follows: The encoder's first tokenized MLP module consists of a PatchEmbed+ShiftedBlock+LayerNorm layer, and its feature map size varies as follows: The second tokenized MLP module consists of PatchEmbed+ShiftedBlock+LayerNorm layers, and the size of its feature map varies as follows:

[0123] After obtaining the feature map through the encoder, the feature map can be further processed through the decoder of the image segmentation model to obtain the block binary map corresponding to each segment of the bone marrow tissue image. Then, based on the coordinates of each block binary map, multiple block binary maps are stitched together to obtain a binary image (Prediction). The contour information of the binary image is then calculated to obtain the bone marrow fibrous region. (The bone marrow fibrous region can be found in [reference needed]). Figure 12 As shown.

[0124] In another embodiment of this application, after segmenting the processed bone marrow tissue image into multiple segmented bone marrow tissue images, the image segmentation model needs to be trained. A flowchart illustrating the training process of the image segmentation model is also provided. Please refer to [link / reference]. Figure 13 As shown, the method includes:

[0125] S201. Preprocess the obtained pathological images of the samples to obtain images of the bone marrow tissue region of the samples.

[0126] S202. The bone marrow tissue region image of the sample is divided into blocks to obtain multiple sample block bone marrow tissue images; each sample block bone marrow tissue image contains the labeled bone marrow fibrous region.

[0127] Specifically, pathological images of the sample can be obtained first. These images can be multiple or a single image. Each image can include at least one bone marrow fibrous region, a bone marrow fat region, and a bone marrow trabecular region. For example, the pathological image can be an argyrophilic section of the bone marrow sample.

[0128] After obtaining the sample pathological image, image processing can be performed on the sample pathological image. First, the sample bone marrow tissue image is converted to BGR format, and then channel separation is performed on the sample bone marrow tissue image to obtain the corresponding B channel image, R channel image and G channel image. Based on the channel image, R channel image and G channel image, the sample binary image is obtained through conversion processing. Then, based on the B channel image, R channel image and sample binary image, edge optimization and contour extraction processing are performed on the sample bone marrow tissue image to obtain the sample bone marrow tissue region image.

[0129] After obtaining the sample bone marrow tissue region image (argyrophilic slice of sample bone marrow tissue), the size of a single argyrophilic slice is very large, with a resolution of possibly 100,000 × 200,000. This size does not meet the input requirements for training an image segmentation model and cannot be directly fed into the model for training. Therefore, it needs to be processed. A single sample bone marrow tissue region image (argyrophilic slice of sample bone marrow tissue) can be cut into multiple sample block bone marrow tissue images (argyrophilic slices of sample bone marrow tissue) of a fixed size. The coordinate position of each sample block bone marrow tissue image can be preserved, facilitating the subsequent stitching of the binary images obtained after model prediction.

[0130] In this process, the labeled bone marrow fibrous regions in each sample segment bone marrow tissue image can be obtained by manual (pathology experts). During the training phase, large sample bone marrow tissue region images are cut into segment bone marrow fibrous region images of a certain size. Since there are many interference terms in the fibrous region, it is necessary to add enough negative samples. In order to ensure that the model is fully trained, each segment bone marrow fibrous region image is equipped with bone marrow fibrous regions during the image cutting process.

[0131] S203. Input the bone marrow tissue image of each sample block into the image segmentation model to be constructed for processing to obtain the sample block binary image corresponding to each sample block bone marrow tissue image.

[0132] S204. The multiple sample block binary images are stitched together to obtain the sample binary image.

[0133] S205. Calculate the contour of the sample binary map to obtain the predicted bone marrow fibrosis region.

[0134] S206. Based on the loss function between the predicted bone marrow fibrous region and the labeled bone marrow fibrous region, the image segmentation model to be constructed is iteratively trained using an iterative algorithm according to minimizing the loss function, so as to obtain the image segmentation model.

[0135] The image segmentation model to be constructed takes a sample block of bone marrow tissue image as input and outputs a sample block binary image. It possesses the ability to binarize the sample block of bone marrow tissue image and can predict the sample binary image using a neural network model. This image segmentation model can be the initial model during iterative training, meaning the model parameters are in their initial state, or it can be the model adjusted in the previous iteration, meaning the model parameters are in an intermediate state. The sample block of bone marrow tissue image can be input into the image segmentation model to obtain the output result, which can include the category result of the sample block of bone marrow tissue image. The category result represents the probability of pixel values ​​of 0 and 255 in the sample block of bone marrow tissue image, with probability values ​​distributed between 0 and 1. Threshold filtering can be used to convert the sample block of bone marrow tissue image into a block binary image. The size of the resulting sample block binary image is the same as the size of the sample block of bone marrow tissue image.

[0136] After obtaining multiple sample block binary images, the coordinates of the corresponding bone marrow tissue image for each sample block binary image can be acquired. These multiple sample block binary images are then stitched together to obtain a sample binary image, which has the same size as the sample bone marrow tissue region image. The contour of the sample binary image is then calculated to obtain the predicted bone marrow fibrous region. A loss function is then constructed based on the predicted and labeled bone marrow fibrous regions. Following the minimization of the loss function, an iterative algorithm is used to iteratively train the image segmentation model to be constructed, resulting in the image segmentation model.

[0137] After obtaining the output of the predicted bone marrow fibrosis region, a loss function can be constructed based on the output and the labeled bone marrow fibrosis region. The image segmentation model to be constructed is then optimized by minimizing the loss function to obtain the image segmentation model. The parameters in the image segmentation model to be constructed are updated according to the difference between the output and the labeled results to achieve the purpose of training the image segmentation model. The labeled results can be the identification results of the bone marrow fibrosis region obtained by manually labeling the sample block bone marrow tissue image.

[0138] Optionally, updating the parameters in the image segmentation model to be constructed may involve updating matrix parameters such as the weight matrix and bias matrix in the image segmentation model to be constructed. The weight matrix and bias matrix mentioned above include, but are not limited to, the matrix parameters in the convolutional layers, feedforward layers, and fully connected layers of the image segmentation model to be constructed.

[0139] When updating the parameters of the image segmentation model to be constructed using the loss function, if the loss function indicates that the model has not converged, the parameters in the model are adjusted to bring it to convergence, thus obtaining the image segmentation model. Convergence of the image segmentation model can be defined as the difference between the output and labeled results being less than a preset threshold, or the rate of change of the difference between the output and labeled results approaching a certain low value. When the calculated loss function is small, or the difference between the calculated loss function and the loss function output from the previous iteration approaches 0, the image segmentation model is considered to have converged, and the image segmentation model is obtained.

[0140] S106. Based on the images of bone marrow tissue regions and bone marrow fibrosis regions, statistically analyze the degree of fibrosis and determine pathological indicators.

[0141] Please see Figure 14 As shown, after identifying the bone marrow tissue region image (silver-stained section of bone marrow tissue) and the bone marrow fibrosis region, the bone marrow tissue region image is in BGR format. Then, the bone marrow tissue region image is divided into blocks according to a fixed size to obtain multiple block bone marrow tissue region images. Each block bone marrow tissue region image and the bone marrow fibrosis region are processed through a fibrosis degree determination model to determine the degree of fibrosis in each block bone marrow tissue region image. Finally, the degree of fibrosis in all block bone marrow tissue region images is statistically analyzed, and the bone marrow fibrosis determination result of this bone marrow tissue region image is output. The bone marrow fibrosis determination result includes the distribution of fibrosis degree and fibrosis pathological indicators.

[0142] Optionally, as one possible implementation of this application, please refer to [link to relevant documentation]. Figure 15 As shown, Figure 15A flowchart illustrating a method for determining the degree of fibrosis in each segment of bone marrow fibrosis region image provided in this application embodiment, the method comprising:

[0143] S301. Obtain an image of the bone marrow tissue region with a known initial fibrosis level.

[0144] S302. Calculate bone marrow fibrosis index information of bone marrow tissue region image based on bone marrow fibrosis region, and correct the initial fibrosis level according to bone marrow fibrosis index information to obtain intermediate fibrosis level; bone marrow fibrosis index information includes bone marrow tissue area, bone marrow fibrosis region area and bone marrow fibrosis region proportion.

[0145] S303. Traverse the bone marrow tissue region image, divide the bone marrow tissue region image into blocks, and obtain multiple block bone marrow tissue region images.

[0146] S304. Calculate the fiber density information of each segment of bone marrow tissue region image.

[0147] S305. Based on fiber density information, construct the relationship between intermediate fiberization layers and fiber density.

[0148] S306. Based on the relationship between intermediate fibrosis levels and fiber density, obtain the target fibrosis level of the segmented bone marrow tissue region image for each fibrosis level.

[0149] It should be noted that the bone marrow tissue images of the known initial fibrosis levels mentioned above can be obtained manually based on the reader's experience in scintigraphy. This initial fibrosis level can include four grades: grade 0 myelofibrosis, grade 1 myelofibrosis, grade 2 myelofibrosis, and grade 3 myelofibrosis. Grade 0 myelofibrosis corresponds to changes in the bone marrow morphology of a normal person. Grade 1 myelofibrosis is characterized by leukocytosis and mild splenomegaly, generally without anemia or thrombocytopenia, and may be asymptomatic; clinically, this is called early or pre-primary primary myelofibrosis. Grade 2 myelofibrosis is characterized by extensive fibrous tissue proliferation visible under a microscope, some of which intertwine into a reticular pattern. Clinical symptoms are pronounced at this stage, including anemia, thrombocytopenia, leukocytosis, and hepatosplenomegaly. Grade 3 myelofibrosis is characterized by worsening clinical symptoms, severe anemia often requiring blood transfusions, leukopenia, recurrent infections, severe hepatosplenomegaly, significant abdominal distension, weight loss, skin and mucous membrane bleeding, and irregular fever.

[0150] Please see Figure 16 As shown, after obtaining a bone marrow tissue region image (bone marrow tissue slice dataset) with a known initial fibrosis level, bone marrow fibrosis index information of the bone marrow tissue region image can be calculated based on the bone marrow fibrosis region. Then, the bone marrow tissue region image is traversed, and the bone marrow tissue region image is divided into blocks to obtain multiple block bone marrow tissue region images. Please refer to [further details omitted]. Figure 17 As shown, Figure 17 Each small square shown represents a segment of bone marrow tissue region in the image.

[0151] The aforementioned bone marrow fibrosis indicators include bone marrow tissue area, bone marrow fibrosis region area, and the proportion of bone marrow fibrosis region. Specifically, for each segment of bone marrow tissue region image, the bone marrow tissue area (Area) can be calculated based on the segment's image. bone The number of bone marrow fibers (Num) is calculated based on the segmented bone marrow fibrous regions. fiber and bone marrow fibrosis area fiber Based on the area of ​​the bone marrow fibrous region and the area of ​​bone marrow tissue, the proportion of the bone marrow fibrous region can be obtained, which can be expressed by the following formula:

[0152]

[0153] Because the initial fibrosis level of the manually determined bone marrow tissue region image may deviate from the actual target fibrosis level due to focusing only on a local area of ​​the image, a primary correction is needed after determining the bone marrow fibrosis index information to obtain an intermediate fibrosis level. During the correction process, adjustments are made based on bone marrow fibrosis level criteria. For example, if the initial fibrosis level is grade 0, and the bone marrow fibrosis index information meets the criteria for grade 1 bone marrow fibrosis, then the corresponding intermediate fibrosis level is determined to be grade 1 bone marrow fibrosis.

[0154] After determining the intermediate fibrosis level of each segment of bone marrow tissue image, the fiber density information of each segment can be calculated. This allows us to obtain the area of ​​the segment's fibrous region and the area of ​​the entire segment. Then, we divide these two areas to obtain the fiber density information for each segment, which can be expressed by the following formula:

[0155]

[0156] Among them, Area fiber_in_patch It refers to the area of ​​segmented bone marrow fibrosis. patch This refers to the area of ​​the segmented bone marrow tissue region in the image. Taking a segment of size 512 as an example, the area of ​​the segmented bone marrow tissue region in the image is 512 * 512 = 262144.

[0157] Then, based on the fiber density information, all segmented bone marrow tissue region images are sorted, and several segmented bone marrow tissue region images with larger fiber density information are selected as the representation units of the bone marrow tissue region image. A relationship between intermediate fibrosis levels and fiber density is constructed, where the intermediate fibrosis level and fibrosis density of each segmented bone marrow tissue region image are known. Based on the relationship between intermediate fibrosis levels and fiber density, each fibrosis level has a multiple of 100 representation units, and each fiber density has a corresponding preset interval range. By integrating all data, the expectation and variance of the Gaussian distribution are calculated, ultimately obtaining the expected interval parameters of the fiber density of the segmented bone marrow tissue region image corresponding to each fibrosis level. This allows the determination of the degree of fibrosis in each segmented bone marrow tissue region image, which is the target fibrosis level.

[0158] After determining the degree of fibrosis (fibrosis level) of each segment of bone marrow tissue image, the degree of fibrosis (fibrosis level) of all segmented bone marrow tissue images is statistically analyzed. That is, in the process of diagnosing myelofibrosis, the degree of fibrosis (fibrosis level) of each segmented bone marrow tissue image is obtained. The degree of fibrosis (fibrosis level) of the bone marrow tissue image (slice) is determined according to the principle of "30% high grade priority". That is, if the number of segments with high fibrosis accounts for more than 30% of the total number of segments, the bone marrow tissue is judged to have a high degree of fibrosis.

[0159] For example, 100 images of bone marrow tissue regions with known fibrosis levels can be obtained, and the fibrosis level of each region can be viewed. For instance, if 30% of the bone marrow tissue region images are of grade II myelofibrosis, 70% of the bone marrow tissue region images are of grade I myelofibrosis, and 10% of the bone marrow tissue region images are of grade zero myelofibrosis, then the fibrosis level (fibrosis level) of the entire bone marrow tissue region image is determined to be grade I.

[0160] For example, please see Figure 18 As shown, using the bone marrow pathology image to be processed as the original argyrophilic bone marrow section, after obtaining the argyrophilic bone marrow section, irrelevant regions of the bone marrow tissue can be filtered to obtain a filtered bone marrow section image. It is then determined whether the filtered bone marrow section image meets the bone marrow preparation standards. The contour coordinates that meet the bone marrow preparation standards are mapped to the corresponding level of the original argyrophilic bone marrow section, thus obtaining the argyrophilic bone marrow sample section. Next, the argyrophilic bone marrow sample section is preprocessed to obtain the bone marrow tissue contour information, resulting in a bone marrow tissue region image. Image processing algorithms are used to locate the bone marrow trabecular bone tissue, bone marrow adipose tissue, and bone marrow fibrous regions. Based on the bone marrow tissue region image and the bone marrow fibrous region, the degree of fibrosis is statistically analyzed to determine pathological indicators.

[0161] The method for determining pathological indicators provided in this application involves acquiring bone marrow pathological images to be processed, preprocessing these images to obtain a set of bone marrow sample images, then processing these images to obtain bone marrow tissue region images. Further image processing is performed on these bone marrow tissue region images to determine the bone marrow adipose tissue region and trabecular bone tissue region. These regions are then removed to obtain a processed bone marrow tissue image. The processed bone marrow tissue image is then input into a trained image segmentation model to obtain a binary image. The contour of the binary image is calculated to obtain the bone marrow fibrous region. Finally, based on the bone marrow tissue region image and the bone marrow fibrous region, the degree of fibrosis is statistically analyzed to determine pathological indicators. This technical solution, by preprocessing the bone marrow pathological images to be processed, can accurately determine the set of bone marrow sample images and the bone marrow tissue region images, providing data guidance for the subsequent determination of the bone marrow adipose tissue region and trabecular bone tissue region. On the other hand, by processing the bone marrow tissue images through a trained image segmentation model, the layout details of the pathological images can be extracted, thereby accurately and quickly obtaining binary images. This allows for precise identification of the bone marrow fibrosis region, enabling more comprehensive and fine-grained statistical analysis of the degree of fibrosis based on the bone marrow tissue region images and the bone marrow fibrosis region. This provides doctors with accurate quantitative results, reduces the instability of pathological indicators caused by subjective human factors, assists doctors in making efficient and accurate judgments, and improves the accuracy of pathological indicator determination.

[0162] It should be noted that although the operations of the method of the present invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0163] on the other hand, Figure 19 This is a schematic diagram of a pathological index determination device provided in an embodiment of this application. This device can be a component within a terminal device, such as… Figure 19 As shown, the device 400 includes:

[0164] The acquisition module 410 is used to acquire bone marrow pathology images to be processed;

[0165] Preprocessing module 420 is used to preprocess the bone marrow pathology images to be processed to obtain a set of bone marrow sample images;

[0166] The first processing module 430 is used to perform image processing on the bone marrow sample image set to obtain bone marrow tissue region images;

[0167] The second processing module 440 is used to perform image processing on the bone marrow tissue region image, determine the bone marrow adipose tissue region and the trabecular bone tissue region, remove the bone marrow adipose tissue region and the trabecular bone tissue region, and obtain the processed bone marrow tissue image.

[0168] The fiber region determination module 450 is used to input the processed bone marrow tissue image into the trained image segmentation model to obtain a binary image, calculate the contour of the binary image, and obtain the bone marrow fiber region.

[0169] The statistical analysis module 460 is used to statistically analyze the degree of fibrosis and determine pathological indicators based on bone marrow tissue area images and bone marrow fibrosis areas.

[0170] Optionally, the preprocessing module 420 mentioned above includes:

[0171] The conversion unit 421 is used to convert the bone marrow pathology image to be processed into a bone marrow image in RGBA format;

[0172] The segmentation processing unit 422 is used to obtain the pixel value of the A channel in the RGBA format bone marrow image, and to segment the bone marrow pathological image to be processed according to the pixel value of the A channel and the preset A channel threshold to obtain multiple bone marrow samples.

[0173] The filtering processing unit 423 is used to filter each bone marrow sample in multiple bone marrow samples according to a preset size pixel threshold to obtain a set of bone marrow sample images.

[0174] Optionally, the first processing module 430 includes:

[0175] The conversion processing unit 431 is used to convert each bone marrow sample image in the bone marrow sample image set into a bone marrow tissue image in BGR format.

[0176] The separation processing unit 432 is used to perform channel separation processing on the bone marrow tissue image to obtain B channel image, R channel image and G channel image, and to obtain a binarized image based on the B channel image, R channel image and G channel image;

[0177] The contour extraction unit 433 is used to perform edge optimization and contour extraction processing on the bone marrow tissue image based on the B channel image, R channel image and binarized image to obtain the bone marrow tissue region image.

[0178] Optionally, the second processing module 440 includes:

[0179] The format conversion unit 441 is used to convert the bone marrow tissue region image into a LAB format bone marrow tissue region image.

[0180] Segmentation unit 442 is used to segment LAB format bone marrow tissue region images into blocks according to a fixed size to obtain multiple segmented images;

[0181] The first determining unit 443 is used to determine the pixel value of each block image in the L color channel for each block image in a plurality of block images, and to determine the fat-like region based on the pixel value of the L color channel and a preset pixel threshold range.

[0182] The second determining unit 444 is used to determine single fat droplet regions and multiple fat aggregation regions based on the morphological characteristics of fat-like regions.

[0183] The third determining unit 445 is used to determine the bone marrow adipose tissue region based on the single fat droplet region and the multiple fat accumulation region.

[0184] Optionally, the fiber region determination module 450 described above is specifically used for:

[0185] Image enhancement processing is performed on the bone marrow tissue region image to obtain an enhanced image;

[0186] The enhanced image was converted into a binary image of the bone marrow mask.

[0187] Calculate the pixel values ​​of all contours in the binary image of the bone marrow mask;

[0188] The binary image of the bone marrow mask is filtered based on a first size threshold, a second size threshold, and pixel values ​​to obtain a complete trabecular bone region and a trabecular bone fragment region; the first size threshold is greater than the second size threshold.

[0189] Based on the intact trabecular bone region and the trabecular bone fragment region, the trabecular bone tissue region is determined.

[0190] Optionally, the fiber region determining module 450 described above is also used for:

[0191] The processed bone marrow tissue image was divided into blocks to obtain multiple segmented bone marrow tissue images.

[0192] Each segment of bone marrow tissue image is input into the encoder of the image segmentation model to obtain a feature map;

[0193] The feature map is passed through the decoder of the image segmentation model to obtain a block-based binary map;

[0194] Multiple segmented binary images are stitched together to obtain a binary image.

[0195] Optionally, the aforementioned statistical analysis module 460 is specifically used for:

[0196] Based on the BGR format, bone marrow tissue region images and bone marrow fibrosis regions are processed using a fibrosis degree determination model to determine the degree of fibrosis in each segment of the bone marrow tissue region image.

[0197] Statistical analysis was performed on the fibrosis degree of all segmented bone marrow tissue regions to determine pathological indicators.

[0198] Optionally, the aforementioned statistical analysis module 460 is also used for:

[0199] Obtain images of bone marrow tissue regions with known initial fibrosis levels;

[0200] Based on the bone marrow fibrosis region, bone marrow fibrosis index information is calculated from the bone marrow tissue region image. The initial fibrosis level is corrected according to the bone marrow fibrosis index information to obtain the intermediate fibrosis level. The bone marrow fibrosis index information includes bone marrow tissue area, bone marrow fibrosis region area, and bone marrow fibrosis region proportion.

[0201] The bone marrow tissue region image is traversed, and the bone marrow tissue region image is divided into blocks to obtain multiple block bone marrow tissue region images.

[0202] Calculate the fiber density information of each segment of bone marrow tissue region image;

[0203] Based on fiber density information, the relationship between intermediate fibrous layers and fiber density is constructed;

[0204] Based on the relationship between intermediate fibrosis levels and fiber density, the target fibrosis level of the segmented bone marrow tissue region image for each fibrosis level is obtained.

[0205] Optionally, the aforementioned statistical analysis module 460 is also used for:

[0206] Calculate the bone marrow tissue area based on segmented bone marrow tissue region images;

[0207] The number of bone marrow fibers and the area of ​​the bone marrow fiber region are calculated based on the bone marrow fibrosis region.

[0208] The proportion of bone marrow fibrous region is obtained based on the area of ​​bone marrow fibrous region and bone marrow tissue area.

[0209] Optionally, the image segmentation model is constructed through the following steps:

[0210] The acquired pathological images of the samples were preprocessed to obtain images of the bone marrow tissue region of the samples;

[0211] The bone marrow tissue region image of the sample is divided into blocks to obtain multiple sample block bone marrow tissue images; each sample block bone marrow tissue image contains labeled bone marrow fibrous regions;

[0212] Each sample block of bone marrow tissue image is input into the image segmentation model to be constructed for processing, resulting in multiple sample block binary images;

[0213] Multiple sample binary images are stitched together to obtain a sample binary image;

[0214] Calculate the contour of the sample binary map to obtain the predicted bone marrow fibrosis region;

[0215] Based on the loss function between the predicted bone marrow fibrous region and the labeled bone marrow fibrous region, an iterative algorithm is used to iteratively train the image segmentation model to be constructed, minimizing the loss function, to obtain the image segmentation model.

[0216] The pathological index determination device provided in this application preprocesses the bone marrow pathological images to accurately determine the bone marrow sample image set and bone marrow tissue region images, providing data guidance for subsequent determination of bone marrow adipose tissue regions and trabecular bone tissue regions. Furthermore, by processing the processed bone marrow tissue images using a trained image segmentation model, the layout details of the pathological images can be extracted, thereby accurately and quickly obtaining binary images. This allows for precise determination of bone marrow fibrosis regions, enabling more comprehensive and fine-grained statistical analysis of the degree of fibrosis based on the bone marrow tissue region images and bone marrow fibrosis regions. This provides doctors with accurate quantitative results, reducing the instability of pathological indicators caused by subjective human factors, assisting doctors in making efficient and accurate judgments, and improving the accuracy of pathological index determination.

[0217] On the other hand, the terminal device provided in the embodiments of this application includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the pathological indicator determination method as described above.

[0218] The following is for reference. Figure 20 , Figure 20 This is a schematic diagram of the computer system structure of the terminal device according to an embodiment of this application.

[0219] like Figure 20 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 602 or programs loaded from storage portion 603 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the system 600. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0220] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.

[0221] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a machine-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 section 603, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined above in the system of this application.

[0222] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0223] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0224] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be housed in a processor; for example, it can be described as: a processor including: an acquisition module, a preprocessing module, a first processing module, a second processing module, a fiber region determination module, and a statistical analysis module. The names of these units or modules do not necessarily limit the specific unit or module itself; for example, the acquisition module can also be described as "for acquiring bone marrow pathology images to be processed".

[0225] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable storage medium stores one or more programs, which, when used by one or more processors, execute the pathological indicator determination method described in this application:

[0226] Acquire bone marrow pathology images to be processed;

[0227] The bone marrow pathology images to be processed are preprocessed to obtain a set of bone marrow sample images;

[0228] The bone marrow sample image set is processed to obtain bone marrow tissue region images;

[0229] Image processing is performed on the bone marrow tissue region image to determine the bone marrow adipose tissue region and the trabecular bone tissue region. The bone marrow adipose tissue region and the trabecular bone tissue region are then removed to obtain the processed bone marrow tissue image.

[0230] The processed bone marrow tissue image is input into a trained image segmentation model to obtain a binary image. The contour of the binary image is calculated to obtain the bone marrow fibrous region.

[0231] Based on the bone marrow tissue region images and the bone marrow fibrosis region, the degree of fibrosis was statistically analyzed to determine pathological indicators.

[0232] In summary, the pathological index determination method, apparatus, equipment, and medium provided in this application involve acquiring bone marrow pathological images to be processed, preprocessing these images to obtain a set of bone marrow sample images, then processing these images to obtain bone marrow tissue region images. Further image processing is performed on these bone marrow tissue region images to determine the bone marrow adipose tissue region and trabecular bone tissue region. These regions are then removed to obtain a processed bone marrow tissue image. The processed bone marrow tissue image is then input into a trained image segmentation model to obtain a binary image. The contour of the binary image is calculated to obtain the bone marrow fibrous region. Finally, based on the bone marrow tissue region image and the bone marrow fibrous region, the degree of fibrosis is statistically analyzed to determine the pathological index. This technical solution, by preprocessing the bone marrow pathological images to be processed, can accurately determine the set of bone marrow sample images and the bone marrow tissue region image, providing data guidance for the subsequent determination of the bone marrow adipose tissue region and trabecular bone tissue region. On the other hand, by processing the bone marrow tissue images through a trained image segmentation model, the layout details of the pathological images can be extracted, thereby accurately and quickly obtaining binary images. This allows for precise identification of the bone marrow fibrosis region, enabling more comprehensive and fine-grained statistical analysis of the degree of fibrosis based on the bone marrow tissue region images and the bone marrow fibrosis region. This provides doctors with accurate quantitative results, reduces the instability of pathological indicators caused by subjective human factors, assists doctors in making efficient and accurate judgments, and improves the accuracy of pathological indicator determination.

[0233] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for determining pathological indicators, characterized in that, The method includes: Acquire bone marrow pathology images to be processed; The bone marrow pathology images to be processed are preprocessed to obtain a set of bone marrow sample images; The bone marrow sample image set is processed to obtain bone marrow tissue region images; Image processing is performed on the bone marrow tissue region image to determine the bone marrow adipose tissue region and the trabecular bone tissue region. The bone marrow adipose tissue region and the trabecular bone tissue region are then removed to obtain the processed bone marrow tissue image. The processed bone marrow tissue image is input into a trained image segmentation model to obtain a binary image. The contour of the binary image is calculated to obtain the bone marrow fibrous region. Obtain bone marrow tissue region images with known initial fibrosis levels; calculate bone marrow fibrosis index information of the bone marrow tissue region images based on the bone marrow fibrosis regions, and correct the initial fibrosis level according to the bone marrow fibrosis index information to obtain intermediate fibrosis levels; the bone marrow fibrosis index information includes bone marrow tissue area, bone marrow fibrosis region area, and bone marrow fibrosis region proportion; traverse the bone marrow tissue region images and divide the bone marrow tissue region images into blocks to obtain multiple block bone marrow tissue region images; calculate the fiber density information of each block bone marrow tissue region image; construct the relationship between the intermediate fibrosis level and fiber density based on the fiber density information; obtain the degree of fibrosis of each fibrosis level block bone marrow tissue region image according to the relationship between the intermediate fibrosis level and fiber density; statistically analyze the degree of fibrosis of all block bone marrow tissue region images to determine pathological indicators.

2. The method according to claim 1, characterized in that, The bone marrow pathology images to be processed are preprocessed to obtain a set of bone marrow sample images, including: The bone marrow pathology image to be processed is converted into a bone marrow image in RGBA format; The pixel values ​​of the A channel in the RGBA format bone marrow image are obtained, and the bone marrow pathological image to be processed is segmented according to the pixel values ​​of the A channel and a preset A channel threshold to obtain multiple bone marrow samples. Each bone marrow sample in a set of multiple bone marrow samples is filtered according to a preset size pixel threshold to obtain a set of bone marrow sample images.

3. The method according to claim 1 or 2, characterized in that, The bone marrow sample image set is processed to obtain bone marrow tissue region images, including: For each bone marrow sample image in the bone marrow sample image set, the bone marrow sample image is converted into a bone marrow tissue image in BGR format; The bone marrow tissue image is subjected to channel separation processing to obtain B-channel image, R-channel image and G-channel image, and a binarized image is obtained based on the B-channel image, R-channel image and G-channel image conversion processing; Based on the B-channel image, the R-channel image, and the binarized image, edge optimization and contour extraction are performed on the bone marrow tissue image to obtain a bone marrow tissue region image.

4. The method according to claim 1, characterized in that, Image processing is performed on the bone marrow tissue region image to determine the bone marrow adipose tissue region, including: The bone marrow tissue region image was converted into a bone marrow tissue region image in LAB format. The LAB format bone marrow tissue region image is divided into blocks of a fixed size to obtain multiple block images; For each of the plurality of segmented images, determine the pixel value of each segmented image in the L color channel, and determine the fat-like region based on the pixel value of the L color channel and a preset pixel threshold range. Based on the morphological characteristics of the aforementioned fat-like regions, single fat droplet regions and multiple fat aggregation regions were identified. Based on the single fat droplet region and the multiple fat accumulation region, the bone marrow adipose tissue region is determined.

5. The method according to claim 1, characterized in that, Image processing is performed on the bone marrow tissue region image to determine the trabecular bone tissue region, including: The bone marrow tissue region image is subjected to image enhancement processing to obtain an enhanced image; The enhanced image is then converted into a binary image of the bone marrow mask. Calculate the pixel values ​​of all contours in the binary image of the bone marrow mask; The binary image of the bone marrow mask is filtered according to a first size threshold, a second size threshold, and the pixel value to obtain a complete trabecular bone region and a trabecular bone fragment region; the first size threshold is greater than the second size threshold. Based on the complete trabecular bone region and the trabecular bone fragment region, the trabecular bone tissue region is determined.

6. The method according to claim 1, characterized in that, The processed bone marrow tissue image is input into a trained image segmentation model to obtain a binary image, including: The processed bone marrow tissue image is divided into blocks to obtain multiple segmented bone marrow tissue images. Each of the segmented bone marrow tissue images is input into the encoder of the image segmentation model to obtain a feature map; The feature map is passed through the decoder of the image segmentation model to obtain a block-based binary map; The multiple segmented binary images are stitched together to obtain a binary image.

7. The method according to claim 1, characterized in that, Based on the bone marrow fibrosis region, bone marrow fibrosis index information of the bone marrow tissue region image is calculated, including: Calculate the bone marrow tissue area based on the segmented bone marrow tissue region image; The number of bone marrow fibers and the area of ​​the bone marrow fiber region are calculated based on the described bone marrow fiber region. The proportion of the bone marrow fibrous region is obtained based on the area of ​​the bone marrow fibrous region and the area of ​​the bone marrow tissue.

8. The method according to claim 6, characterized in that, The image segmentation model is constructed through the following steps: The acquired pathological images of the samples were preprocessed to obtain images of the bone marrow tissue region of the samples; The bone marrow tissue region image of the sample is segmented to obtain multiple sample segmented bone marrow tissue images; Each of the sample block bone marrow tissue images contains labeled bone marrow fibrous regions; Each sample block bone marrow tissue image is input into the image segmentation model to be constructed for processing, resulting in multiple sample block binary images. The multiple sample block binary images are stitched together to obtain a sample binary image; The contour of the sample binary image is calculated to obtain the predicted bone marrow fibrosis region; Based on the loss function between the predicted bone marrow fibrosis region and the labeled bone marrow fibrosis region, the image segmentation model to be constructed is iteratively trained using an iterative algorithm according to minimizing the loss function, thereby obtaining the image segmentation model.

9. A device for determining pathological indicators, characterized in that, The device includes: The acquisition module is used to acquire bone marrow pathology images to be processed; The preprocessing module is used to preprocess the bone marrow pathology images to be processed to obtain a set of bone marrow sample images; The first processing module is used to perform image processing on the bone marrow sample image set to obtain bone marrow tissue region images; The second processing module is used to perform image processing on the bone marrow tissue region image, determine the bone marrow adipose tissue region and the trabecular bone tissue region, remove the bone marrow adipose tissue region and the trabecular bone tissue region, and obtain the processed bone marrow tissue image. The fiber region determination module is used to input the processed bone marrow tissue image into a trained image segmentation model to obtain a binary image, calculate the contour of the binary image, and obtain the bone marrow fiber region. The statistical analysis module is used to acquire bone marrow tissue region images with known initial fibrosis levels; calculate bone marrow fibrosis index information of the bone marrow tissue region images based on the bone marrow fibrosis regions; correct the initial fibrosis level according to the bone marrow fibrosis index information to obtain intermediate fibrosis levels; the bone marrow fibrosis index information includes bone marrow tissue area, bone marrow fibrosis region area, and bone marrow fibrosis region proportion; traverse the bone marrow tissue region images, segment the bone marrow tissue region images to obtain multiple segmented bone marrow tissue region images; calculate the fiber density information of each segmented bone marrow tissue region image; construct the relationship between the intermediate fibrosis level and fiber density based on the fiber density information; obtain the degree of fibrosis of each fibrosis level segmented bone marrow tissue region image according to the relationship between the intermediate fibrosis level and fiber density; statistically analyze the degree of fibrosis of all segmented bone marrow tissue region images to determine pathological indicators.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the pathological indicator determination method as described in any one of claims 1-8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the pathological indicator determination method as described in any one of claims 1-8.

Images