| abstract |
The invention relates to the technical fields of computer vision and medical image analysis, and specifically discloses a method, system, equipment and storage medium for classifying images of tumors in the esophagus-gastric junction. The specific steps of the classification method are: S1, processing the tumor medical image of the esophagogastric junction of the subject, and obtaining a three-dimensional ROI area image of the tumor medical image lesion; S2, extracting the image group in the three-dimensional ROI area image. S3, input the value of the radiomics feature extracted in step S2 into the scoring prediction model, and calculate the image score of the three-dimensional ROI region image; S4, carry out qualitative analysis on the image score obtained in step S3, predict the tumor medical image image type. The classification method of the invention can predict whether the esophagogastric junction tumor belongs to adenocarcinoma or squamous cell carcinoma before operation, and the predicted AUC can reach more than 0.8, which has high accuracy; and has the advantage of non-invasiveness before operation. |