A CMR image segmentation and classification system

Abstract

The invention discloses a CMR image segmentation and classification system comprising an image preprocessing module and an image segmentation and classification module. The image preprocessing moduleis used for obtaining and cutting an original CMR image and obtaining a target area image; the image segmentation and classification module is a trained convolutional neural network and comprises an encoder, a decoder and an image post-processing module; wherein the encoder comprises a convolutional layer, a spatial pyramid pooling layer and a first softmax classifier which are connected in sequence; a shallow feature map of the target area image is output by the convolutional layer; the first softmax classifier outputs a deep feature map and a classification result of the target area image; the decoder is used for fusing the shallow feature map and the deep feature map to obtain a segmentation probability response map; and the image post-processing module is used for obtaining a segmentation mask of the target area image according to the segmentation probability response map. According to the technical scheme provided by the invention, automatic and accurate target region segmentationand image classification can be carried out on the CMR image, and doctors are assisted to carry out disease diagnosis.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a CMR image segmentation and classification system. Background technique [0002] Image semantic segmentation (Semantic image Segmentation) and image classification (ImageClassification) are classic research topics in the field of artificial intelligence and computer vision. The semantic segmentation of images aims to use computer algorithms to automatically classify the regions where different objects are located in the image pixel by pixel to form a segmentation mask. Image semantic segmentation requires that the segmentation mask obtained by the segmentation algorithm can retain accurate edge details and the same resolution size as the original image, and at the same time correctly classify different objects in the image. Different from the semantic segmentation of images, image classification is to classify the entire image, and find the most likely label in the exi...

AI Technical Summary

Problems solved by technology

Obviously, this manual delineation process is not only cumbersome and inefficient, but also the delineation accuracy largely depends on the doctor's personal experience and professionalism

Images