METHOD FOR DETECTING AND CLASSIFYING LESION AREA IN CLINICAL IMAGE

The present disclosure discloses a method for detecting and classifying lesion areas in clinical images. The method includes: preprocessing original clinical images to obtain a global image data set; inputting the obtained global image data set into a region-based convolutional neural network (RCNN) model for training; marking each of the detected possible lesion areas with a rectangular box, cutting rectangular box areas to obtain lesion areas, and performing normalisation and data enhancement on the lesion areas to obtain a local image data set containing the lesion areas; and inputting the global image data set and the local image data set containing the lesion areas into a two-stream convolutional neural network (CNN) for classification as dual modalities. The method of the present disclosure uses multi-modal information of the clinical image more excellently, and combines global information of the original image and local information of the lesion area image, thereby increasing accuracy of classification.