CN-122025017-A - Tumor interventional therapy preoperative planning system based on artificial intelligence image recognition

CN122025017ACN 122025017 ACN122025017 ACN 122025017ACN-122025017-A

Abstract

The application relates to the technical field of edge segmentation, in particular to a tumor interventional therapy preoperative planning system based on artificial intelligent image recognition, which is used for preliminarily screening out an infiltration edge region based on the asymmetric gray distribution condition of a sliding window in a CT image and the texture confusion condition in an MRI image; the DSA vascular data are further fused, the region of interest is screened out through analyzing the position distribution condition of the vascular region of the infiltration edge region, then the curvature mutation condition of the tumor associated vascular segment is analyzed on the basis of the region of interest to generate a vascular abnormality index, finally the infiltration index and the vascular characteristic are fused to construct a vascular-infiltration coupling index, the edge intensity map is corrected according to the vascular abnormality index, and boundary optimization based on risk classification is realized, so that the accuracy of the obtained tumor region edge structure curve is higher, and the accuracy of tumor region segmentation according to the tumor region edge structure curve is improved.

Inventors

ZHANG YONG
YUAN HONGXUN
GAO KUN
ZHOU CHUANGUO
WANG JIANFENG
YANG XIANGYU
LIU KEYUN

Assignees

北京大学国际医院
首都医科大学附属北京朝阳医院

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. A tumor interventional therapy pre-operative planning system based on artificial intelligence image recognition, the system comprising: the data acquisition preprocessing module is used for positioning the same tumor ROI region in the tumor CT image, the tumor MRI image and the tumor DSA image, traversing the tumor ROI region to determine all sliding windows, and acquiring a blood vessel region in the tumor DSA image based on angiography; the first determining module is used for determining the comprehensive infiltration index of each sliding window based on the asymmetric gray level distribution condition of each sliding window in the tumor CT image and the disordered gray level distribution condition of each sliding window in the tumor MRI image; The second determining module is used for determining a corresponding infiltration activity characteristic value according to the position distribution condition of the blood vessel region of each infiltration edge region in the tumor DSA image, screening out an interesting infiltration region according to the infiltration activity characteristic value, and determining a corresponding blood vessel abnormality index according to the bending change condition of the blood vessel region in the interesting infiltration region; The region segmentation module is used for determining a corresponding vascular infiltration coupling index according to the comprehensive infiltration index, the infiltration activity characteristic value and the vascular abnormality index of each interesting infiltration region, carrying out edge intensity map correction on the tumor ROI region according to the vascular infiltration coupling index, determining a corresponding tumor region edge structure curve, and carrying out tumor region segmentation according to the tumor region edge structure curve.
2. The preoperative planning system for tumor interventional therapy based on artificial intelligence image recognition according to claim 1, wherein the process of obtaining the comprehensive infiltration index comprises the following steps: in the tumor CT image, determining a corresponding cell proliferation activity index according to the integral size of the gray value in each sliding window and the uniform distribution of the gray value; Normalizing the numerical entropy of gray values of all pixel points in each sliding window in the MRI image, and determining the corresponding tissue component mixing degree; and determining the comprehensive infiltration index of each sliding window according to the product of the negative correlation mapping value of the cell proliferation activity index and the tissue component hybridization degree.
3. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 2, wherein the cell proliferation activity index obtaining process comprises: In the tumor CT image, the average value of gray values of all pixels in each sliding window is normalized to determine an average cell density coefficient, negative correlation normalization is performed according to the skewness of the gray values of all pixels in each sliding window to determine a density distribution symmetry coefficient, and the cell proliferation activity index of each sliding window is determined according to the product between the average cell density coefficient and the density distribution symmetry coefficient.
4. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 1, wherein the acquiring process of the infiltrated edge area comprises: and taking the area corresponding to the sliding window corresponding to the comprehensive infiltration index larger than the preset index threshold as an infiltration edge area.
5. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 1, wherein the process of acquiring the characteristic value of infiltration activity comprises the following steps: in the tumor DSA image, skeletonizing a blood vessel region in each infiltration edge region, and determining all skeleton pixel points on a corresponding blood vessel skeleton; determining corresponding basic vascular infiltration probability according to the spatial position approach condition of each reference pixel point and all skeleton pixel points; And determining a corresponding characteristic value of the infiltration activity according to the average value of the basic vascular infiltration probabilities of all the reference pixel points in each infiltration edge region.
6. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 5, wherein the process for obtaining the basal vascular infiltration probability comprises: And carrying out negative correlation mapping on the minimum Euclidean distance between each reference pixel point and all skeleton pixel points, and determining the basic vascular infiltration probability of each reference pixel point.
7. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 1, wherein the acquisition process of the interesting infiltration region comprises: And taking the infiltration edge area corresponding to the infiltration activity characteristic value larger than the preset activity threshold value as the concerned infiltration area.
8. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 5, wherein the process of obtaining the vascular abnormality index comprises: determining a corresponding curvature contrast value according to the average value of the curvatures of all adjacent skeleton pixel points of each skeleton pixel point on the vascular skeleton of each concerned infiltration area; Normalizing the average value of the local curvature deviation values of all skeleton pixel points on the vascular skeleton of each concerned infiltration region, and determining the vascular abnormality index of each concerned infiltration region.
9. The pre-interventional tumor planning system based on artificial intelligence image recognition according to claim 1, wherein the process of obtaining the vascular infiltration coupling index comprises: normalizing the product among the comprehensive infiltration index, the infiltration activity characteristic value and the vascular abnormality index of each concerned infiltration region, and determining the corresponding vascular infiltration coupling index.
10. The pre-tumor interventional therapy planning system based on artificial intelligence image recognition according to claim 1, wherein the process of obtaining the tumor region edge structure curve comprises the following steps: In a tumor CT image, arranging the edge intensities of all pixel points in the concerned infiltration region in a sequence from large to small to determine a corresponding initial edge intensity sequence, upwardly rounding the product between the positive correlation mapping value of the vascular infiltration coupling index and the preset threshold number in the initial edge intensity sequence, and determining the edge selection number of each concerned infiltration region; And performing B spline curve fitting on pixel points corresponding to the edge intensities of the front edge selected number in the edge intensity sequence, and determining an edge structure curve of the tumor area.

Description

Tumor interventional therapy preoperative planning system based on artificial intelligence image recognition Technical Field The invention relates to the technical field of edge segmentation, in particular to a tumor interventional therapy preoperative planning system based on artificial intelligent image recognition. Background The tumor area is usually positioned based on a U-Net++ segmentation model on the basis of CT images, then a segmentation edge line is determined through a canny edge algorithm, and a tumor area is segmented through the segmentation edge line, but the accurate edge of the tumor area is usually affected by infiltration to cause boundary blurring, so that the canny edge algorithm can only segment strong edges when the segmentation edge line is segmented, but neglect the infiltration edge under the influence of cell dispersion-edema-blood vessel coupling, namely the recognition accuracy of the canny edge algorithm on the infiltration edge in the tumor area in the prior art is lower, and the accuracy of the segmentation of the tumor area is affected. Disclosure of Invention In order to solve the technical problem that the recognition accuracy of the infiltration edge is low when a canny edge algorithm is used for dividing a tumor region, the application aims to provide a tumor interventional therapy preoperative planning system based on artificial intelligent image recognition, and the adopted technical scheme is as follows: The first aspect of the application provides a tumor interventional therapy preoperative planning system based on artificial intelligence image recognition, which comprises: the data acquisition preprocessing module is used for positioning the same tumor ROI region in the tumor CT image, the tumor MRI image and the tumor DSA image, traversing the tumor ROI region to determine all sliding windows, and acquiring a blood vessel region in the tumor DSA image based on angiography; the first determining module is used for determining the comprehensive infiltration index of each sliding window based on the asymmetric gray level distribution condition of each sliding window in the tumor CT image and the disordered gray level distribution condition of each sliding window in the tumor MRI image; The second determining module is used for determining a corresponding infiltration activity characteristic value according to the position distribution condition of the blood vessel region of each infiltration edge region in the tumor DSA image, screening out an interesting infiltration region according to the infiltration activity characteristic value, and determining a corresponding blood vessel abnormality index according to the bending change condition of the blood vessel region in the interesting infiltration region; The region segmentation module is used for determining a corresponding vascular infiltration coupling index according to the comprehensive infiltration index, the infiltration activity characteristic value and the vascular abnormality index of each interesting infiltration region, carrying out edge intensity map correction on the tumor ROI region according to the vascular infiltration coupling index, determining a corresponding tumor region edge structure curve, and carrying out tumor region segmentation according to the tumor region edge structure curve. Further, the process for obtaining the comprehensive infiltration index comprises the following steps: in the tumor CT image, determining a corresponding cell proliferation activity index according to the integral size of the gray value in each sliding window and the uniform distribution of the gray value; Normalizing the numerical entropy of gray values of all pixel points in each sliding window in the MRI image, and determining the corresponding tissue component mixing degree; and determining the comprehensive infiltration index of each sliding window according to the product of the negative correlation mapping value of the cell proliferation activity index and the tissue component hybridization degree. Further, the process for obtaining the cell proliferation activity index comprises the following steps: In the tumor CT image, the average value of gray values of all pixels in each sliding window is normalized to determine an average cell density coefficient, negative correlation normalization is performed according to the skewness of the gray values of all pixels in each sliding window to determine a density distribution symmetry coefficient, and the cell proliferation activity index of each sliding window is determined according to the product between the average cell density coefficient and the density distribution symmetry coefficient. Further, the obtaining process of the infiltrated edge area includes: and taking the area corresponding to the sliding window corresponding to the comprehensive infiltration index larger than the preset index threshold as an infiltration edge area. Further, the obtaining process