Pancreatic operation key anatomical region segmentation method based on CT image

Abstract

The invention relates to the technical field of image processing, in particular to a pancreatic operation key anatomical region segmentation method based on a CT image, and the method comprises the following steps: obtaining a CT arterial phase image and a CT venous phase image; preprocessing the CT arterial phase image and the CT venous phase image in a window width and window level adjustment and nonlinear filtering processing mode; and through combination of a TotalSegmentor segmentation model, an nnUNet segmentation framework and a region growing algorithm, key blood vessels such as a pancreatic duodenal artery disjunction region and a PV-SMV axis are segmented from the preprocessed CT vein phase image and the preprocessed CT vein phase image, and peripheral adipose tissues are obtained. According to the method, key blood vessels such as a pancreatic duodenal artery dissection region and a PV-SMV axis are automatically segmented by combining an nnNet segmentation framework to optimize a TotalSegmentor technology, and a surrounding fat gap is obtained, so that the problem of subjective deviation caused by traditional manual ROI sketching is solved, key anatomical regions such as a GDA region and a PV-SMV region are accurately positioned, and the accuracy of the segmentation of the pancreatic duodenal artery dissection region is improved. And a more stable space frame which better meets the surgical requirements is provided for image omics feature extraction.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method for segmenting key anatomical regions in pancreatic surgery based on CT images. Background Art

[0002] Pancreaticoduodenectomy (PD) is the standard surgical procedure for treating lesions of the pancreatic head and periampullary region. Resection involves the dense vascular and neural plexus within the mesopancreatic region (e.g., SMA, SMV / PV, etc.), and therefore is considered one of the most complex and challenging general surgical procedures. Inflammation in the mesopancreatic region (e.g., chronic pancreatitis, fibrosis after neoadjuvant therapy) replaces the loose perivascular spaces with dense fibrous tissue, leading to a loss of normal perivascular anatomical layers and a significant increase in the incidence of "difficult PD."

[0003] Preoperative contrast-enhanced CT is the primary method for assessing inflammation in the surgical area, but technical limitations limit the clinical value of conventional CT findings. Although CT images clearly demonstrate the anatomy of the large vessels surrounding the pancreas, their low soft tissue resolution can prevent the identification of early microscopic inflammatory features (such as microcirculatory impairment), leading to underestimation of the extent of intraoperative inflammation in the pancreatic mesentery. Furthermore, existing imaging assessment strategies often focus on morphological changes in the "tumor-vascular contact area," overlooking the impact of inflammation on the mechanical properties of local tissues, a key factor in predicting the difficulty of vascular dissection.

[0004] Radiomics extracts texture, morphology, and functional features from medical images at high throughput, transforming microscopic tissue heterogeneity (such as the spatial distribution of inflammation-related fibrosis) into computable high-dimensional data. This data has the potential to be used to construct clinically meaningful inflammation prediction models. Therefore, the construction of inflammation prediction models relies on the radiomics features of key pancreatic surgical regions. Therefore, it is paramount to accurately segment these key pancreatic surgical regions in CT images for radiomics feature extraction. Currently, manual delineation methods used for segmentation of key pancreatic surgical regions are subject to subjective bias, and automated delineation using artificial intelligence algorithms cannot accurately locate key anatomical regions such as the gastroduodenal artery (GDA) and portal vein-superior mesenteric vein (PV-SMV). This means that the framework cannot provide a more stable and surgically appropriate spatial segmentation framework for radiomics feature extraction. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for segmenting key anatomical regions for pancreatic surgery based on CT images, so as to solve the technical problem in the existing technology of accurately locating various key anatomical regions such as the gastroduodenal artery (GDA) and portal vein-superior mesenteric vein (PV-SMV), that is, it is unable to provide a more stable and more surgically suitable spatial segmentation framework for imaging genomics feature extraction.

[0006] In order to solve the above technical problems, the present invention specifically provides the following technical solutions: A method for segmenting key anatomical regions for pancreatic surgery based on CT images comprises the following steps: Obtain CT arterial phase images and CT venous phase images; Preprocessing the CT arterial phase image and the CT venous phase image by adjusting the window width and window position and performing nonlinear filtering processing; By combining the TotalSegmentator segmentation model, the nnUNet segmentation framework, and the region growing algorithm, the connective tissue ROI 1 of the LPD surgical area, the peripancreatic connective tissue ROI 2, the connective tissue ROI 3 around the common hepatic artery and its main branches, the connective tissue ROI 4 around the superior mesenteric artery, the connective tissue ROI 5 around the SMV-PV axis, and the connective tissue ROI 6 on the right side of the SMV-PV axis were segmented in the preprocessed CT venous phase images and the preprocessed CT arterial phase images, respectively.

[0007] As a preferred embodiment of the present invention, the segmentation method of the LPD surgical area connective tissue ROI 1 includes: On the pre-processed CT venous phase images, the TotalSegmentator segmentation model was used to segment the important anatomical structures during LPD surgery and determine the cropping frame including the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branches. On the CT venous phase image within the cropping box, the abdominal visceral fat is locally segmented using the nnUNet segmentation framework; On the pre-processed CT venous phase images, global segmentation of abdominal visceral fat was performed using the nnUnet segmentation framework; The local segmentation result of the abdominal visceral fat is spatially registered with the global segmentation result of the abdominal visceral fat and a mask operation is performed to obtain the LPD operation area connective tissue ROI 1.

[0008] As a preferred embodiment of the present invention, the segmentation method of the peripancreatic connective tissue ROI 2 includes: On the pre-processed CT venous phase images, the pancreas contour was segmented using the TotalSegmentator segmentation model; Taking the overall outline of the pancreas as a reference, the surrounding three-dimensional space was uniformly expanded by 10 mm to generate the peripancreatic connective tissue ROI2; The characterization of peripancreatic connective tissue ROI 2 was optimized by setting the CT value range from -150HU to 60HU.

[0009] As a preferred embodiment of the present invention, the segmentation method of the connective tissue ROI 3 surrounding the common hepatic artery and its main branches and the connective tissue ROI 4 surrounding the superior mesenteric artery includes: On preprocessed CT arterial phase images, the lower boundary was manually determined, with the lower edge of the 12th thoracic vertebra (i.e., the horizontal plane at the origin of the abdominal aorta) as the upper boundary. The left and right and anterior and posterior boundaries were determined based on the overall contour of the pancreas. A cubic region of interest was constructed encompassing the common hepatic artery, its major branches, and the superior mesenteric artery. The major branches of the common hepatic artery included the proper hepatic artery and the origin of the gastroduodenal artery. The pixel values ​​of the vertebral body, pancreas, and aorta within the cubic region of interest were set to zero. The high-density characteristics of blood vessels in CT images (CT value range: 150HU to 300HU) were used to trace the arterial tissue from the beginning of the abdominal aorta using a region growing algorithm. The region growing algorithm first identifies the first major branch of the anterior abdominal aorta: the celiac trunk, followed by the right branches of the celiac trunk: the common hepatic artery, the origin of the gastroduodenal artery, and the origin of the proper hepatic artery. Then, the region growing algorithm identified the second major branch of the abdominal aorta, the superior mesenteric artery, 1 to 2 cm below the celiac trunk; The common hepatic artery, the origin of the gastroduodenal artery, the origin of the proper hepatic artery, and the superior mesenteric artery were three-dimensionally isotropically expanded with an expansion radius of 10 mm to obtain connective tissue ROI 3 around the common hepatic artery, the origin of the gastroduodenal artery, and the origin of the proper hepatic artery, as well as connective tissue ROI 4 around the superior mesenteric artery.

[0010] As a preferred embodiment of the present invention, the segmentation method of the connective tissue ROI 5 around the SMV-PV axis and the connective tissue ROI 6 on the right side of the SMV-PV axis includes: On the pre-processed CT venous phase images, an initial three-dimensional region of interest (ROI) was constructed, including the vein, portal vein, and splenic vein, with the upper edge of the third lumbar vertebra (i.e., the plane where the horizontal part of the duodenum is located) as the lower boundary and the portal vein bifurcation as the upper boundary. The pixel values ​​of the vertebral body, pancreas, superior mesenteric vein, portal vein, and splenic vein in the initial 3D region of interest were set to zero. Taking advantage of the high density of vascular tissue in CT venous phase images (CT value range: 150HU to 300HU), a region growing algorithm was used to track venous tissue starting from the portal vein bifurcation to identify the superior mesenteric vein, portal vein, and splenic vein. The splenic vein region was clipped 300 pixels to the left of the central axis of the CT venous phase image to obtain the SMV-PV axis; The SMV-PV axis was three-dimensionally isotropically expanded with an expansion radius of 10 mm to obtain ROI 5 of the connective tissue around the SMV-PV axis; The SMV-PV axis was taken as the center, and the right half of the SMV-PV axis was identified using a region growing algorithm with a growth radius of 5 mm to obtain the connective tissue ROI 6 on the right side of the SMV-PV axis.

[0011] As a preferred embodiment of the present invention, the method of spatially registering the local segmentation result of abdominal visceral fat with the global segmentation result of abdominal visceral fat and performing mask calculation includes: Performing spatial registration on the global segmentation result of the abdominal visceral fat and the clipping frame to obtain the global segmentation result of the abdominal visceral fat in the clipping frame coordinate system; Extracting the global segmentation result of the abdominal visceral fat within the clipping frame from the global segmentation result of the abdominal visceral fat within the clipping frame coordinate system; Real-time acquisition of the segmentation accuracy performance of the nnUNet segmentation framework for local segmentation of abdominal visceral fat, and real-time acquisition of the segmentation accuracy performance of the nnUNet segmentation framework for global segmentation of abdominal visceral fat; Based on the segmentation accuracy of the local segmentation and the segmentation accuracy of the global segmentation, a mask operation is performed on the global segmentation result of the abdominal visceral fat within the cropping frame and the local segmentation result of the abdominal visceral fat in a weighted fusion manner to obtain the LPD surgical area connective tissue ROI 1; The expression of the mask operation is: ; Where, is the connective tissue of the LPD operation area, is the global segmentation result of the abdominal visceral fat within the cropping box, This is the local segmentation result of abdominal visceral fat. is the segmentation accuracy of the global segmentation, is the segmentation accuracy of local segmentation.

[0012] As a preferred embodiment of the present invention, the arterial tissue and venous tissue tracking method of the region growing algorithm includes: Step 1: Use the pancreas mask frame to crop as the venous whole seed, and use the abdominal aorta and the upper and lower ends of the pyramid L1-2 to crop the abdominal aorta as the arterial whole seed; Step 2: Apply morphological expansion operation to the vein whole seed and artery whole seed to perform transverse expansion of vein tissue and artery tissue, and do not expand the two ends longitudinally. The expression of the morphological expansion is: ; ; Where Ω is the area after transverse expansion, x is the point in the area after expansion, y is the point in the center line, b is the structural element point of morphological expansion, c(t) is the center line of artery and vein, To exclude the areas near the two end points of arteries and veins, only The constant term for dilating the centerline within the range, r is the dilation radius, C′ is the centerline portion remaining after removing the two end points of the artery and vein, Br is the morphological structure element of radius r, and ⊕ is the Minkowski operator; Step 3: Determine the vascular space constraint conditions and select the growth areas of venous tissue and arterial tissue based on the vascular space constraint conditions. The expression of the vascular space constraint conditions is: ; Where Ω_vessel is the spatial region suitable for blood vessel growth, D(x) is the characteristic attribute of point x in the expanded region, is a predefined direction vector, is the threshold of the growth direction, is the grayscale value of point x in the expanded area, and They are the minimum and maximum values ​​of the preset range of grayscale values ​​respectively; Step 4: Iterate steps 2 and 3 to obtain the common hepatic artery, the origin of the gastroduodenal artery and the origin of the proper hepatic artery, the superior mesenteric artery, the superior mesenteric vein, the portal vein, and the splenic vein.

[0013] As a preferred solution of the present invention, the region growing algorithm identifies the right half of the SMV-PV axis with a growth radius of 5 mm and adopts a layered expansion method based on distance transformation plus direction restriction. The layered expansion method based on distance transformation plus direction restriction includes: The binary mask of the SMV-PV axis is inverted, and the distance transform algorithm is used to obtain the distance map of the pixels outside the SMV-PV axis to the nearest SMV-PV axis boundary; On the distance map, a given 5 mm threshold is used to first filter out all pixels that are no more than 5 mm away from the SMV-PV axis boundary. The coordinates of the pixels are compared with the coordinates of their nearest blood vessel boundary points. Pixels that meet the 5 mm threshold are only retained in the area to the right of the SMV-PV axis. Among them, if the SMV-PV axis is curved and the definition of "right side" needs to rely on the local normal, the center line of the SMV-PV axis is first extracted and the tangent vector and normal vector at each point are calculated. Then, among the candidate points whose distance meets the 5 mm threshold, the candidate points within the "right normal" range of that point are judged and retained. The expansion area within 5 mm to the right of the SMV-PV axis is obtained as the connective tissue ROI 6 on the right side of the SMV-PV axis.

[0014] As a preferred solution of the present invention, the gold standard for segmentation of ROI 1 to ROI 6 is obtained by manual segmentation by a radiologist with professional experience in abdominal imaging.

[0015] As a preferred embodiment of the present invention, the segmentation accuracy is quantified by the difference between the global segmentation result of the abdominal visceral fat within the cropping frame and the local segmentation result of the abdominal visceral fat and the segmentation gold standard of ROI 1. The expression for quantifying the segmentation accuracy is: ; ; Where, is the global segmentation result within the cropping box obtained by the nnUNet segmentation framework on the i-th image sample at the current moment, is the local segmentation result obtained by the nnUNet segmentation framework on the i-th image sample at the current moment, is the gold standard for segmentation of ROI 1 on the i-th image sample, is the mean square loss formula, and m is the total number of image samples.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention optimizes the TotalSegmentator technology by combining the nnNet segmentation framework to achieve automatic segmentation of key blood vessels such as the pancreaticoduodenal artery disconnection area and the PV-SMV axis and obtain the surrounding fat gap. It not only solves the subjective bias problem of traditional manual ROI delineation, but also provides a more stable and more surgically appropriate spatial framework for imaging feature extraction by accurately locating key anatomical areas such as the GDA area and the PV-SMV area. In addition, this nnNet segmentation framework combined with the TotalSegmentator technology to form a hierarchical modeling strategy for the key area segmentation model has for the first time shifted the surgeon's visual evaluation criteria from the overall visual field clarity to the local anatomical resolution, providing a new technical path for the accurate prediction of inflammation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other implementation drawings based on the provided drawings without inventive effort.

[0018] Figure 1 A flow chart of a method for segmenting key anatomical regions for pancreatic surgery based on CT images provided in an embodiment of the present invention; Figure 2 Flowchart of the LPD surgical area connective tissue ROI 1 segmentation method provided in an embodiment of the present invention; Figure 3 A result diagram of the region of interest segmentation provided by an embodiment of the present invention; Figure 4 A schematic diagram of a collected data set provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, the present invention provides a method for segmenting key anatomical regions for pancreatic surgery based on CT images, comprising the following steps: Obtain CT arterial phase images and CT venous phase images; CT arterial phase images and CT venous phase images were preprocessed using window width and window level adjustment and nonlinear filtering. The window width was set from 350 Hounsfield units (HU) to 400 HU, and the window level was set from 40 HU to 50 HU. By combining the TotalSegmentator segmentation model, the nnUNet segmentation framework, and the region growing algorithm, the connective tissue ROI 1 of the LPD surgical area, the peripancreatic connective tissue ROI 2, the connective tissue ROI 3 around the common hepatic artery and its main branches, the connective tissue ROI 4 around the superior mesenteric artery, the connective tissue ROI 5 around the SMV-PV axis (superior mesenteric vein-portal vein axis), and the connective tissue ROI 6 on the right side of the SMV-PV axis were segmented in the preprocessed CT venous phase images and the preprocessed CT arterial phase images, respectively.

[0021] To achieve precise segmentation of key anatomical regions for pancreatic surgery, the present invention constructs an artificial intelligence model to automatically segment the connective tissue ROI 1 of the LPD surgical area, the peripancreatic connective tissue ROI 2, the connective tissue ROI 3 around the common hepatic artery and its main branches (proper hepatic artery and gastroduodenal artery), the connective tissue ROI 4 around the superior mesenteric artery, the connective tissue ROI 5 around the SMV-PV axis, and the connective tissue ROI 6 on the right side of the SMV-PV axis. This model automatically segments key blood vessels such as the pancreaticoduodenal artery disconnection area and the PV-SMV axis and obtains the surrounding fat gap. This not only solves the subjective bias problem of traditional manual ROI delineation, but also provides a more stable and surgically tailored spatial framework for radiomics feature extraction by accurately locating key anatomical regions such as the GDA area (gastroduodenal artery) and PV-SMV area.

[0022] In constructing the artificial intelligence model, the present invention adopts a hierarchical modeling strategy to combine the TotalSegmentator segmentation model and the nnUNet segmentation framework to construct a segmentation structure of key anatomical regions for pancreatic surgery. The TotalSegmentator segmentation model in the segmentation structure pre-segments multiple organs including target areas such as the pancreas and large blood vessels, and then uses the nnUNet segmentation framework to perform fine segmentation of other key areas such as the entire abdominal fat on the basis of the coarse segmentation, thereby improving the segmentation accuracy while retaining the information of multiple anatomical structures.

[0023] The present invention also adds a region growing algorithm to the pancreatic surgery key anatomical region segmentation structure composed of the TotalSegmentator segmentation model and the nnUNet segmentation framework, thereby realizing the comprehensive use of the TotalSegmentator segmentation model, the nnUNet segmentation framework and the region growing algorithm to complete the segmentation of various pancreatic surgery key anatomical regions, including the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branch structures, and surrounding fat spaces.

[0024] Among them, the TotalSegmentator segmentation model can identify the blood vessel contour based on the threshold and simplify the blood vessel image into a skeleton line. Each pixel on this line represents the center point of a blood vessel cross section. After the simplification is completed, the end point or the next bifurcation point of the blood vessel is found by identifying the topological structure of the skeleton line, thereby segmenting the blood vessel into multiple independent branches. Therefore, the present invention can segment the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branch structures through the TotalSegmentator segmentation model, and realize the accurate segmentation of key blood vessels such as the pancreatic duodenal artery disconnection area and the PV-SMV axis.

[0025] The nnUNet segmentation framework is an adaptive deep learning framework designed specifically for medical image segmentation. Through an automated configuration process, it can achieve the optimal performance of the basic UNet architecture in segmenting tumors or extracting organ contours in different data sets. Therefore, the present invention applies it to the clipping frame obtained by the TotalSegmentator segmentation model, which includes the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branch structures. The LPD surgical area connective tissue ROI 1, peripancreatic connective tissue ROI 2, connective tissue ROI 3 around the common hepatic artery and its main branches, connective tissue ROI 4 around the superior mesenteric artery, connective tissue ROI 5 around the SMV-PV axis, and connective tissue ROI 6 on the right side of the SMV-PV axis can be segmented, that is, other key areas such as abdominal visceral fat can be segmented.

[0026] Therefore, the hierarchical modeling strategy combining the TotalSegmentator segmentation model and the nnUNet segmentation framework can segment key blood vessels such as the pancreaticoduodenal artery disconnection area and the PV-SMV axis in CT images, while also segmenting the surrounding fat gaps, achieving the fusion of coarse and fine segmentation, and improving segmentation accuracy while retaining multiple anatomical structure information.

[0027] The region growing algorithm completes image segmentation based on regional features. Specifically, starting from a given or automatically selected seed point, it repeatedly detects pixels that meet the conditions in the region based on similarities in grayscale, curvature, and texture, and gradually merges adjacent pixels that meet the conditions to obtain the target region.

[0028] This paper incorporates a region growing algorithm into the segmentation of abdominal visceral fat (connective tissue) and other key regions within the nnUNet segmentation framework. This algorithm uses the cropped segmentation mask of blood vessels and the spine as the overall "seed," without specifying a single seed point. Instead, all voxels within the mask are considered the starting region. This seed-based expansion and constrained screening approach, rather than a seed-by-seed region growing algorithm, improves processing efficiency while maintaining a more integrated expanded region and ensuring that the expanded region does not arbitrarily "intrude" into tissue that does not conform to HU values ​​or morphological characteristics.

[0029] In the present invention, the nnUNet segmentation framework performs segmentation of abdominal visceral tissue on a cropping frame including the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branch structures. It is a local image segmentation based on CT images, which realizes refined segmentation focusing on key areas. However, compared with the segmentation of the global image, the local image may inevitably lose some image information, resulting in impaired segmentation accuracy. In order to compensate for the loss of segmentation accuracy caused by the loss of this part of information, the present invention also performs a segmentation of abdominal visceral tissue on the original CT image with the nnUNet segmentation framework. It is a global image segmentation based on the CT image, and the global image segmentation result is added to the local image segmentation result through spatial alignment and mask operation, which is equivalent to adding the original image information to the local segmentation to reduce information loss, thereby avoiding the loss of segmentation accuracy caused by information loss, and further improving the segmentation accuracy of the hierarchical modeling strategy.

[0030] The present invention adopts a hierarchical modeling strategy to combine the TotalSegmentator segmentation model (from TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CTImages. Radiology: Artificial Intelligence. https: / / doi.org / 10.1148 / ryai.230024.) and the nnUNet segmentation framework to construct the segmentation structure of key anatomical regions for pancreatic surgery, such as Figure 4 As shown in Figure 2, the dataset included: 112 patients in the non-inflammation group and 91 patients in the inflammation group were randomly assigned to the training set and validation set at a ratio of 7:3. 44 patients in the mild inflammation group and 47 patients in the severe inflammation group were randomly assigned to the training set and validation set at a ratio of 7:3.

[0031] This paper uses a combination of the TotalSegmentator segmentation model, the nnUNet segmentation framework, and the region growing algorithm to segment key anatomical regions for pancreatic surgery, including the pancreas, the SMV-PV axis, the celiac trunk, the common hepatic artery and its main branches, and the surrounding fat space, as follows: like Figure 2 As shown in Figure 1, the segmentation method of connective tissue ROI 1 in the LPD surgical area includes: On the pre-processed CT venous phase images, the TotalSegmentator segmentation model was used to segment the important anatomical structures during LPD surgery and determine the cropping frame including the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branches. On the CT venous phase image within the cropping box, the abdominal visceral fat is locally segmented using the nnUNet segmentation framework; On the pre-processed CT venous phase images, global segmentation of abdominal visceral fat was performed using the nnUnet segmentation framework; The local segmentation results of abdominal visceral fat are spatially registered with the global segmentation results of abdominal visceral fat and mask operations are performed to obtain the connective tissue ROI 1 of the LPD operation area (recorded as Original when predicting the presence or absence of inflammation, and as V RegionFat when predicting mild inflammation and severe inflammation, as shown in Figure 2). Figure 3 (as shown in a to c).

[0032] The segmentation method of peripancreatic connective tissue ROI 2 includes: On the pre-processed CT venous phase images, the pancreas contour was segmented using the TotalSegmentator segmentation model; Taking the overall outline of the pancreas as the reference, the surrounding three-dimensional space was uniformly expanded by 10 mm to generate the peripancreatic connective tissue ROI 2 (denoted as V Pancreas Original, as shown in Figure 2). Figure 3 (shown in d~f); By setting the CT value range to -150HU to 60HU, the CT value range is set to -150HU to 60HU when performing peripancreatic connective tissue segmentation. This CT value range covers normal connective tissue (CT value range: -150HU to -30HU) and inflammatory connective tissue (CT value range: -30HU to 60HU), thereby more comprehensively reflecting the pathophysiological state of the peripancreatic connective tissue and further optimizing the characterization of peripancreatic connective tissue ROI 2.

[0033] The segmentation methods for connective tissue ROI 3 surrounding the common hepatic artery and its main branches and connective tissue ROI 4 surrounding the superior mesenteric artery include: On preprocessed CT arterial phase images, the lower boundary was manually determined, with the lower edge of the 12th thoracic vertebra (i.e., the horizontal plane at the origin of the abdominal aorta) as the upper boundary. The left and right and anterior and posterior boundaries were determined based on the overall contour of the pancreas. A cubic region of interest was constructed encompassing the common hepatic artery, its major branches, and the superior mesenteric artery. The major branches of the common hepatic artery included the proper hepatic artery and the origin of the gastroduodenal artery. The pixel values ​​of the vertebral body, pancreas, and aorta within the cubic region of interest were set to zero. The high-density characteristics of blood vessels in CT images (CT value range: 150HU to 300HU) were used to trace the arterial tissue from the beginning of the abdominal aorta using a region growing algorithm. The region growing algorithm first identifies the first major branch of the anterior abdominal aorta: the celiac trunk, followed by the right branches of the celiac trunk: the common hepatic artery, the origin of the gastroduodenal artery, and the origin of the proper hepatic artery. Then, the region growing algorithm identified the second major branch of the abdominal aorta, the superior mesenteric artery, 1 to 2 cm below the celiac trunk; The common hepatic artery, the origin of the gastroduodenal artery, the origin of the proper hepatic artery, and the superior mesenteric artery were expanded three-dimensionally with an expansion radius of 10 mm to obtain ROI 3 of the connective tissue around the common hepatic artery, the origin of the gastroduodenal artery, and the origin of the proper hepatic artery (denoted as A1 Original, as shown in Figure 2). Figure 3 (g) and the connective tissue ROI 4 around the superior mesenteric artery (denoted as A2 Original, as shown in Figure 3 The present invention uses the above method to accurately identify the connective tissue surrounding the main blood supply artery of the pancreatic head in the pancreatic mesentery.

[0034] The segmentation methods for connective tissue ROI 5 around the SMV-PV axis and connective tissue ROI 6 on the right side of the SMV-PV axis include: On the pre-processed CT venous phase images, an initial three-dimensional region of interest (ROI) was constructed, including the vein, portal vein, and splenic vein, with the upper edge of the third lumbar vertebra (i.e., the plane where the horizontal part of the duodenum is located) as the lower boundary and the portal vein bifurcation as the upper boundary. The pixel values ​​of the vertebral body, pancreas, superior mesenteric vein, portal vein, and splenic vein in the initial 3D region of interest were set to zero. Taking advantage of the high density of vascular tissue in CT venous phase images (CT value range: 150HU to 300HU), a region growing algorithm was used to track venous tissue starting from the portal vein bifurcation to identify the superior mesenteric vein, portal vein, and splenic vein. The splenic vein region was cropped 300 pixels to the left of the central axis of the CT venous phase image to obtain the SMV-PV axis. The splenic vein was not the target vessel of this study, so it was cropped 300 pixels to the left of the central axis of the image (the original image resolution was 512 × 512). The SMV-PV axis was obtained after the splenic vein region was cropped. The SMV-PV axis was expanded three-dimensionally and isotropically with an expansion radius of 10 mm to obtain the connective tissue ROI 5 around the SMV-PV axis (denoted as V1 Original, as shown in Figure 5). Figure 3 (as shown in g); The SMV-PV axis was taken as the center, and the right half of the SMV-PV axis was identified using the region growing algorithm with a growth radius of 5 mm to obtain the connective tissue ROI 6 on the right side of the SMV-PV axis (denoted as V1rightoriginal).

[0035] In the present invention, the nnUNet segmentation framework performs segmentation of abdominal visceral tissue on a cropping frame including the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branch structures. It is a local image segmentation based on CT images, which realizes refined segmentation focusing on key areas. However, compared with the segmentation of the global image, the local image may inevitably lose some image information, resulting in impaired segmentation accuracy. In order to compensate for the loss of segmentation accuracy caused by the loss of this part of information, the present invention also performs a segmentation of abdominal visceral tissue on the original CT image with the nnUNet segmentation framework. It is a global image segmentation based on the CT image. The global image segmentation result is added to the local image segmentation result through spatial registration and mask operation, which is equivalent to adding the original image information to the local segmentation to reduce information loss, thereby avoiding the loss of segmentation accuracy caused by information loss, and further improving the segmentation accuracy of the hierarchical modeling strategy, as follows: The method of spatially registering the local segmentation result of abdominal visceral fat with the global segmentation result of abdominal visceral fat and performing mask operation includes: Performing spatial registration on the global segmentation result of the abdominal visceral fat and the clipping frame to obtain the global segmentation result of the abdominal visceral fat in the clipping frame coordinate system; Extracting the global segmentation result of the abdominal visceral fat within the clipping frame from the global segmentation result of the abdominal visceral fat within the clipping frame coordinate system; Real-time acquisition of the segmentation accuracy performance of the nnUNet segmentation framework for local segmentation of abdominal visceral fat, and real-time acquisition of the segmentation accuracy performance of the nnUNet segmentation framework for global segmentation of abdominal visceral fat; Based on the segmentation accuracy of local segmentation and the segmentation accuracy of global segmentation, the global segmentation results of the abdominal visceral fat within the cropping frame and the local segmentation results of the abdominal visceral fat are subjected to mask operation in a weighted fusion manner to obtain the LPD surgical area connective tissue ROI 1; The expression of mask operation is: ; Where, It is the connective tissue of the LPD operation area. is the global segmentation result of the abdominal visceral fat within the cropping box, This is the local segmentation result of abdominal visceral fat. is the segmentation accuracy of the global segmentation, is the segmentation accuracy of local segmentation.

[0036] The present invention adopts a weighted fusion method in the mask operation, and introduces the nnUNet segmentation framework to perform real-time segmentation accuracy of local segmentation and global segmentation, that is, to judge the global segmentation model performance and local segmentation model performance of the nnUNet segmentation framework at the current moment. The higher the performance, the higher the confidence of the segmentation result. Therefore, after normalization, as the global segmentation result and the weight of the local segmentation result, the segmentation accuracy lost by the local segmentation can be supplemented by global segmentation, so as to achieve the purpose of "adding the global image segmentation result to the local image segmentation result, which is equivalent to adding the original image information to the local segmentation to reduce information loss, thereby avoiding the loss of segmentation accuracy caused by information loss, and further improving the segmentation accuracy of the hierarchical modeling strategy."

[0037] The arterial and venous tissue tracking method of the region growing algorithm includes the following steps: first, cropping the pancreas mask frame as the overall venous seed, and cropping the abdominal aorta with the upper and lower ends of the abdominal aorta and cone L1-2 as the overall arterial seed; Step 2: Apply morphological expansion operation to the vein and artery whole seeds to perform lateral expansion of vein and artery tissues, and do not expand the two ends longitudinally. The expression of morphological expansion is: ; ; Where Ω is the area after transverse expansion, x is the point in the area after expansion, y is the point in the center line, b is the structural element point of morphological expansion, c(t) is the center line of artery and vein, To exclude the areas near the two end points of arteries and veins, only The constant term for dilating the centerline within the range, r is the dilation radius or the size of the structural element, that is, the distance the centerline is laterally "expanded", C′ is the remaining centerline after removing the two end points of the artery and vein, Br is the morphological structural element of radius r, and ⊕ is the Minkowski operator; Centerline parameterization: Assume that the centerline of the blood vessel can be represented by one (or more) curves. For a continuous curve c(t), Represents the parameters from the starting point to the end point. The starting point is c(0) and the end point is c(1). In order to "remove the endpoint", only consider , and ignore the endpoints corresponding to t=0 and t=1 ( It can be set to a very small positive number as needed, or it can be set to 0 directly for special processing during expansion later).

[0038] Define the dilation radius (or structural element): The vessel is usually dilated in the "transverse" direction, i.e., the area on the curve with a distance r. For example, in the two-dimensional case, a disk Br (a circle with a radius of r) is used as the structural element.

[0039] Step 3: Determine the vascular space constraints and select the growth areas of venous and arterial tissues based on the vascular space constraints. The expression of the vascular space constraints is: ; Where Ω_vessel is the spatial region suitable for blood vessel growth, D(x) is the characteristic attribute of point x in the expanded region, is a predefined direction vector, is the threshold of the growth direction, is the grayscale value of point x in the expanded area, and They are the minimum and maximum values ​​of the preset range of grayscale values ​​respectively; The vascular spatial constraint conditions of the present invention are obtained based on anatomical prior knowledge. Based on the relative position clipping, the vascular growth spatial constraint is also obtained based on anatomical prior knowledge, which includes two parts of prior: the first is the direction of the blood vessel, and the second is the HU value intensity of the blood vessel.

[0040] Ω_vessel represents a spatial region suitable for vascular growth; x is the growth point to be evaluated. The first constraint, D(x)·v_prior≥θ_dir, ensures that the relationship between the characteristic attribute D(x) of point x and the predefined direction vector v_prior meets the growth direction threshold θ_dir, thereby screening for suitable growth directions. The second constraint, HU_min≤I(x)≤HU_max, limits the grayscale value I(x) of point x to a reasonable range, ensuring that only tissue that meets vascular characteristics is selected. These constraints enable effective screening of growth regions for venous and arterial tissue.

[0041] Step 4: Iterate steps 2 and 3 to obtain the common hepatic artery, the origin of the gastroduodenal artery and the origin of the proper hepatic artery, the superior mesenteric artery, the superior mesenteric vein, the portal vein, and the splenic vein.

[0042] The present invention uses the initial mask clipped by vascular and vertebral positioning as the overall "seed" without specifying a single seed point, and considers all voxels currently within the mask as the starting area at the same time; Morphological dilation: Dilates the mask, expanding it outward based on the pixel / voxel neighborhood. Before dilation, the mask can be preprocessed as needed (such as denoising, thinning, closing small holes, etc.) to improve the accuracy of subsequent growth.

[0043] Constraint screening: During the expansion process, each candidate voxel in the expansion area is checked to see if it meets the specified constraints: HU value range: If the CT value of a voxel is outside the reasonable range of blood vessels or target tissues, it will not be incorporated into the new mask; Direction / neighborhood features: can be combined with vascular geometric direction, gradient information or other morphological features to filter out areas that expand in unreasonable directions.

[0044] Only voxels that satisfy all constraints are incorporated into the final “outward-expanded” vessel segmentation mask.

[0045] Iterative or multi-step dilation: Usually, dilation is not performed just once. Instead, dilation and screening are performed using a smaller dilation kernel, followed by several rounds of iterations to gradually obtain more accurate boundaries. If the vessel edge is relatively smooth or the HU value varies significantly, only a few iterations are often required to reach a stable state.

[0046] This whole-seed-based expansion and constrained screening approach, rather than a seed-by-seed region growing algorithm, can improve processing efficiency while maintaining a more integrated nature of the expanded region and ensuring that the expanded range does not arbitrarily "invade" tissues that do not meet HU values ​​or morphological characteristics.

[0047] The region growing algorithm uses a layered expansion method based on distance transformation and direction restriction to identify the right half of the SMV-PV axis with a growth radius of 5 mm. The layered expansion method based on distance transformation and direction restriction includes: The binary mask of the SMV-PV axis is inverted, and the distance transform algorithm is used to obtain the distance map of the pixels outside the SMV-PV axis to the nearest SMV-PV axis boundary; On the distance map, a given 5 mm threshold is used to first filter out all pixels that are no more than 5 mm away from the SMV-PV axis boundary. The coordinates of the pixels are compared with the coordinates of their nearest blood vessel boundary points. Pixels within the 5 mm threshold are retained only in the area with a larger x-coordinate (to the right of the SMV-PV axis). Among them, if the SMV-PV axis is curved and the definition of "right side" needs to rely on the local normal, the center line of the SMV-PV axis is first extracted and the tangent vector and normal vector at each point are calculated. Then, among the candidate points whose distance meets the 5 mm threshold, the candidate points within the "right normal" range of that point are judged and retained. The expansion area within 5 mm to the right of the SMV-PV axis is obtained as the connective tissue ROI 6 on the right side of the SMV-PV axis.

[0048] To preserve only the expansion effect on the "right side" of the vessel, the present invention requires combining coordinate information or local normals to impose directional constraints. If only the "positive x-axis" in the image coordinate system is considered, the coordinates of the pixel and its nearest vessel boundary point can be compared. For pixels within a distance of 5 mm, only those in the area with a "larger x-coordinate" are retained. If the vessel is curved and the definition of "right side" relies on the local normal, the vessel centerline must first be extracted and the tangent vector and normal vector at each point calculated. Then, for candidate points whose distance meets the threshold, further determination is made as to whether they are within the "right normal" at that location. This ultimately results in an expansion area within 5 mm to the right of the vessel.

[0049] The local normal is the vector perpendicular to the tangent vector at a specific point on a curve or surface. When dealing with the SMV-PV axis, the local normal is used to determine the left-right orientation at a curve bend. Mathematically, given a point P on a curve, its tangent vector T represents the orientation of the curve at that point, while the normal vector N is the vector perpendicular to the tangent direction, indicating the direction towards the inside or outside of the curve.

[0050] The right normal is a special case of the local normal, typically defined by the orientation of the tangent vector. On the SMV-PV axis, the right normal refers to the region formed by the normal vector to the right of the tangent vector. By calculating the normal vector for each point, it is possible to determine which candidate points lie in the right relative region of the SMV-PV axis. This is crucial for accurately identifying and expanding perivascular connective tissue. The gold standard for segmentation of ROIs 1 to 6 was manually performed by a radiologist with expertise in abdominal imaging.

[0051] The real-time segmentation accuracy of the nnUNet segmentation framework for local and global segmentation in the present invention is measured by a training set obtained from collected case samples. That is, the error between the local and global segmentation results obtained by the nnUNet segmentation framework in the training set and the segmentation gold standard of ROI 1 obtained by manual segmentation by a radiologist with professional experience in abdominal imaging is calculated. The higher the error, the lower the segmentation performance and the lower the confidence of the segmentation result.

[0052] The segmentation accuracy is quantified by the difference between the global segmentation result of the abdominal visceral fat within the cropping box and the local segmentation result of the abdominal visceral fat and the segmentation gold standard of ROI 1. The expression for quantification of segmentation accuracy is: ; ; Where, is the global segmentation result within the cropping box obtained by the nnUNet segmentation framework on the i-th image sample at the current moment, is the local segmentation result obtained by the nnUNet segmentation framework on the i-th image sample at the current moment, is the gold standard for segmentation of ROI 1 on the i-th image sample, is the mean square loss formula, and m is the total number of image samples, which is consistent with the number of cases in the training set.

[0053] In summary, combining the TotalSegmentator segmentation model with nnUNet can achieve a more comprehensive, accurate, and automated workflow for pancreatic key region segmentation, mainly in the following three aspects: 1. Collaboration between coarse and fine segmentation of multiple structures: The TotalSegmentator segmentation model pre-segments multiple organs, including target areas such as the pancreas and large blood vessels. nnUNet then performs fine segmentation of other key areas, such as total abdominal fat, based on this coarse segmentation. This improves segmentation accuracy while preserving information from multiple anatomical structures.

[0054] 2. Automated configuration and high scalability: nnUNet has built-in automated data preprocessing, network structure selection, and hyperparameter tuning, which can significantly reduce model development and debugging time. The TotalSegmentator segmentation model provides a stable multi-organ segmentation framework and pre-trained weights, eliminating the need for tedious manual setup and facilitating the introduction of new anatomical targets or the expansion and tuning of existing target structures.

[0055] 3. Enhanced robustness and generalization: nnUNet effectively addresses data discrepancies across multiple centers and scanning protocols through mechanisms such as automatic normalization, data augmentation, and post-processing. The TotalSegmentator segmentation model, based on large-scale, multi-structured training data, exhibits strong adaptability to diverse clinical images and reduces the impact of domain shifts. This provides more accurate surgical planning and tissue identification for both clinical and scientific research.

[0056] The present invention optimizes the TotalSegmentator technology by combining the nnNet segmentation framework to achieve automatic segmentation of key blood vessels such as the pancreaticoduodenal artery disconnection area and the PV-SMV axis and obtain the surrounding fat gap. It not only solves the subjective bias problem of traditional manual ROI delineation, but also provides a more stable and more surgically appropriate spatial framework for imaging genomics feature extraction by accurately locating key anatomical areas such as the GDA area and the PV-SMV area. In addition, this nnNet segmentation framework combined with the TotalSegmentator technology constitutes a hierarchical modeling strategy for the key area segmentation model. For the first time, the surgeon's visual evaluation standard has shifted from overall visual field clarity to local anatomical resolution, providing a new technical path for the accurate prediction of inflammation.

[0057] The above embodiments are merely exemplary embodiments of the present application and are not intended to limit the scope of the present application. The scope of protection of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and scope of protection of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the scope of protection of the present application.

Claims

1. A method for segmenting key anatomical regions in pancreatic surgery based on CT images, characterized in that: The following steps are involved: Obtain CT arterial phase images and CT venous phase images; Preprocessing the CT arterial phase image and the CT venous phase image by adjusting the window width and window position and performing nonlinear filtering processing; By combining the TotalSegmentator segmentation model, the nnUNet segmentation framework, and the region growing algorithm, the connective tissue ROI 1 of the LPD surgical area, the peripancreatic connective tissue ROI 2, the connective tissue ROI 3 around the common hepatic artery and its main branches, the connective tissue ROI 4 around the superior mesenteric artery, the connective tissue ROI 5 around the SMV-PV axis, and the connective tissue ROI 6 on the right side of the SMV-PV axis were segmented in the preprocessed CT venous phase images and preprocessed CT arterial phase images, respectively.

2. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The segmentation method of the LPD surgical area connective tissue ROI 1 includes: On the pre-processed CT venous phase images, the TotalSegmentator segmentation model was used to segment the important anatomical structures during LPD surgery and determine the cropping frame including the pancreas, SMV-PV axis, celiac trunk, common hepatic artery and its main branches. On the CT venous phase image within the cropping box, the abdominal visceral fat is locally segmented using the nnUNet segmentation framework; On the pre-processed CT venous phase images, global segmentation of abdominal visceral fat was performed using the nnUnet segmentation framework; The local segmentation result of the abdominal visceral fat is spatially registered and masked with the global segmentation result of the abdominal visceral fat to obtain the LPD operation area connective tissue ROI 1.

3. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The segmentation method of the peripancreatic connective tissue ROI 2 includes: On the pre-processed CT venous phase images, the pancreas contour was segmented using the TotalSegmentator segmentation model; Taking the overall contour of the pancreas as the reference, the area was uniformly expanded 10 mm into the surrounding three-dimensional space to generate the peripancreatic connective tissue ROI 2; The characterization of peripancreatic connective tissue ROI 2 was optimized by setting the CT value range from -150HU to 60HU.

4. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The segmentation method of the connective tissue ROI 3 surrounding the common hepatic artery and its main branches and the connective tissue ROI 4 surrounding the superior mesenteric artery includes: On preprocessed CT arterial phase images, the lower boundary was manually determined, with the lower edge of the 12th thoracic vertebra as the upper boundary. The left and right and anterior and posterior boundaries were determined based on the overall contour of the pancreas. A cubic region of interest was constructed encompassing the common hepatic artery, its major branches, and the superior mesenteric artery. The major branches of the common hepatic artery included the proper hepatic artery and the origin of the gastroduodenal artery. The pixel values ​​of the vertebral body, pancreas and aorta in the cubic region of interest were set to zero. The high density of blood vessels in CT images was used to trace the arterial tissue from the beginning of the abdominal aorta using the region growing algorithm. The region growing algorithm first identifies the first major branch of the anterior abdominal aorta: the celiac trunk, followed by the right branches of the celiac trunk: the common hepatic artery, the origin of the gastroduodenal artery, and the origin of the proper hepatic artery. Then, the region growing algorithm identified the second major branch of the abdominal aorta, the superior mesenteric artery, 1 to 2 cm below the celiac trunk; The common hepatic artery, the origin of the gastroduodenal artery, the origin of the proper hepatic artery, and the superior mesenteric artery were three-dimensionally isotropically expanded with an expansion radius of 10 mm to obtain connective tissue ROI 3 around the common hepatic artery, the origin of the gastroduodenal artery, and the origin of the proper hepatic artery, as well as connective tissue ROI 4 around the superior mesenteric artery.

5. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The segmentation method of the connective tissue ROI 5 around the SMV-PV axis and the connective tissue ROI 6 on the right side of the SMV-PV axis includes: On the preprocessed CT venous phase images, an initial three-dimensional region of interest (ROI) including the vein, portal vein, and splenic vein was constructed with the upper edge of the third lumbar vertebra as the lower boundary and the portal vein bifurcation as the upper boundary. The pixel values ​​of the vertebral body, pancreas, superior mesenteric vein, portal vein, and splenic vein in the initial 3D region of interest were set to zero. Taking advantage of the high density of vascular tissue in CT venous phase images, a region growing algorithm was used to track venous tissue starting from the portal vein bifurcation to identify the superior mesenteric vein, portal vein, and splenic vein. The splenic vein region was clipped 300 pixels to the left of the central axis of the CT venous phase image to obtain the SMV-PV axis; The SMV-PV axis was three-dimensionally isotropically expanded with an expansion radius of 10 mm to obtain the connective tissue ROI5 around the SMV-PV axis; The SMV-PV axis was taken as the center, and the right half of the SMV-PV axis was identified using a region growing algorithm with a growth radius of 5 mm to obtain the connective tissue ROI 6 on the right side of the SMV-PV axis.

6. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The method of performing spatial registration and mask operation on the local segmentation result of abdominal visceral fat and the global segmentation result of abdominal visceral fat includes: Performing spatial registration on the global segmentation result of the abdominal visceral fat and the clipping frame to obtain the global segmentation result of the abdominal visceral fat in the clipping frame coordinate system; Extracting the global segmentation result of the abdominal visceral fat within the clipping frame from the global segmentation result of the abdominal visceral fat within the clipping frame coordinate system; Real-time acquisition of the segmentation accuracy performance of the nnUNet segmentation framework for local segmentation of abdominal visceral fat, and real-time acquisition of the segmentation accuracy performance of the nnUNet segmentation framework for global segmentation of abdominal visceral fat; Based on the segmentation accuracy of the local segmentation and the segmentation accuracy of the global segmentation, a mask operation is performed on the global segmentation result of the abdominal visceral fat within the cropping frame and the local segmentation result of the abdominal visceral fat in a weighted fusion manner to obtain the LPD surgical area connective tissue ROI 1; The expression of the mask operation is: ； Where, is the connective tissue of the LPD operation area, is the global segmentation result of the abdominal visceral fat within the cropping box, This is the local segmentation result of abdominal visceral fat. is the segmentation accuracy of the global segmentation, is the segmentation accuracy of local segmentation.

7. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The arterial tissue and venous tissue tracking method of the region growing algorithm includes: Step 1: Use the pancreas mask frame to crop as the venous whole seed, and use the abdominal aorta and the upper and lower ends of the pyramid L1-2 to crop the abdominal aorta as the arterial whole seed; Step 2: Apply morphological expansion operation to the vein whole seed and artery whole seed to perform transverse expansion of vein tissue and artery tissue, and do not expand the two ends longitudinally. The expression of the morphological expansion is: ； ； Where Ω is the area after transverse expansion, x is the point in the area after expansion, y is the point in the center line, b is the structural element point of morphological expansion, c(t) is the center line of artery and vein, To exclude the areas near the two end points of arteries and veins, only The constant term for dilating the centerline within the range, r is the dilation radius, C′ is the centerline portion remaining after removing the two end points of the artery and vein, Br is the morphological structure element of radius r, and ⊕ is the Minkowski operator; Step 3: Determine the vascular space constraint conditions and select the growth areas of venous tissue and arterial tissue based on the vascular space constraint conditions. The expression of the vascular space constraint conditions is: ； Where Ω_vessel is the spatial region suitable for blood vessel growth, D(x) is the characteristic attribute of point x in the expanded region, is a predefined direction vector, is the threshold of the growth direction, is the grayscale value of point x in the expanded area, and They are the minimum and maximum values ​​of the preset range of grayscale values ​​respectively; Step 4: Iterate steps 2 and 3 to obtain the common hepatic artery, the origin of the gastroduodenal artery and the origin of the proper hepatic artery, the superior mesenteric artery, the superior mesenteric vein, the portal vein, and the splenic vein.

8. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 5, characterized in that: The region growing algorithm identifies the right half of the SMV-PV axis with a growth radius of 5 mm using a layered expansion method based on distance transformation plus direction restriction. The layered expansion method based on distance transformation plus direction restriction includes: The binary mask of the SMV-PV axis is inverted, and the distance transform algorithm is used to obtain the distance map of the pixels outside the SMV-PV axis to the nearest SMV-PV axis boundary; On the distance map, a given 5 mm threshold is used to first filter out all pixels that are no more than 5 mm away from the SMV-PV axis boundary. The coordinates of the pixels are compared with the coordinates of their nearest blood vessel boundary points. Pixels that meet the 5 mm threshold are retained only in the area to the right of the SMV-PV axis. If the SMV-PV axis is curved and the definition of "right side" depends on the local normal, the centerline of the SMV-PV axis is first extracted and the tangent vector and normal vector at each point are calculated. Then, among the candidate points that meet the 5 mm distance threshold, the candidate points within the right normal range of that point are identified and retained. The expansion area within 5 mm to the right of the SMV-PV axis is obtained as the connective tissue ROI 6 on the right side of the SMV-PV axis.

9. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 1, characterized in that: The gold standard for segmentation of ROI 1 to ROI 6 was obtained by manual segmentation by a radiologist with professional experience in abdominal imaging.

10. The method for segmenting key anatomical regions for pancreatic surgery based on CT images according to claim 6, characterized in that: The segmentation accuracy is quantified by the difference between the global segmentation result of the abdominal visceral fat within the cropping frame and the local segmentation result of the abdominal visceral fat and the segmentation gold standard of ROI 1. The expression for quantifying the segmentation accuracy is: ； ； Where, is the global segmentation result within the cropping box obtained by the nnUNet segmentation framework on the i-th image sample at the current moment, is the local segmentation result obtained by the nnUNet segmentation framework on the i-th image sample at the current moment, is the gold standard for segmentation of ROI 1 on the i-th image sample, is the mean square loss formula, and m is the total number of image samples.

Images