Cardiac catheterization image recognition and evaluation method
A cardiac catheterization image recognition and evaluation method is disclosed. The first deep learning algorithm is used to conduct an object recognition process on the cardiac catheterization image to obtain the vessel object image. The image processing process is conducted to the cardiac catheterization image to obtain the vessel location image. The vessel object image and the vessel location image are combined to obtain the vessel contour image. The vessel type judging process is conducted to the vessel contour image to determine the type of vessel in the cardiac catheterization image. The second deep learning algorithm is used on the vessel contour image to detect the vessel occlusion location and to judge the vessel occlusion rate. Based on the type of vessel and the vessel occlusion rate at the vessel occlusion location, the cardiac catheterization image is evaluated to obtain the SYNTAX Score.
1 . A cardiac catheterization image recognition and evaluation method comprising:
obtaining a cardiac catheterization image of a patient by an input device,
storing the cardiac catheterization image in a memory of a computing device,
accessing the memory by a processor of the computing device
conducting an object recognition process on the cardiac catheterization image by a first deep learning algorithm to obtain a vessel object image of the cardiac catheterization image;
conducting an image processing process to the cardiac catheterization image to obtain a vessel location image of the cardiac catheterization image;
combining the vessel object image and the vessel location image to obtain a vessel contour image of the cardiac catheterization image;
conducting a vessel type judging process to the vessel contour image to determine a vessel type in the cardiac catheterization image;
detecting a vessel occlusion location by using a second deep learning algorithm on the vessel contour image and judging a vessel occlusion rate of the vessel occlusion location by an occlusion rate calculation process;
and evaluating the cardiac catheterization image to obtain a cardiac anatomy score based on the vessel type and the vessel occlusion rate at the vessel occlusion location, wherein the cardiac anatomy score is provided to medical personnel by an output device for judging vascular occlusion status.
2 . The cardiac catheterization image recognition and evaluation method of claim 1 , wherein the first deep learning algorithm comprises U-Net model or YOLO v4 model, the first deep learning algorithm is used to recognize vessels, bones, catheter, diaphragm and heart of the cardiac catheterization image.
3 . The cardiac catheterization image recognition and evaluation method of claim 1 , wherein the image processing process comprises image preprocessing, vessel segmentation and voting mechanism.
4 . The cardiac catheterization image recognition and evaluation method of claim 3 , wherein the image preprocessing comprises contrast enhancement, image cutting and image smoothing.
5 . The cardiac catheterization image recognition and evaluation method of claim 3 , wherein the vessel segmentation comprises ridge detection, skeleton extraction, noise removal and superpixel segmentation.
6 . The cardiac catheterization image recognition and evaluation method of claim 5 , wherein the voting mechanism comprises overlapping a mask of the ridge detection with the superpixel segmentation and determining the vessel location image by weight voting.
7 . The cardiac catheterization image recognition and evaluation method of claim 1 , wherein the vessel type comprises right dominance and left dominance.
8 . The cardiac catheterization image recognition and evaluation method of claim 1 , wherein the second deep learning algorithm comprises Mask RCNN model and Bi-LSTM model, and after marking the vessel contour image, the vessel occlusion location is predicted by the second deep learning algorithm.
9 . The cardiac catheterization image recognition and evaluation method of claim 1 , wherein a reference point of the vessel contour image is found by tangent line and normal vector line, corresponding points on both sides of vessel are obtained by a dynamic programming algorithm, and vessel width is calculated by the corresponding points, so as to calculates the vessel occlusion rate.