Lumbar intervertebral disc MRI image classification method based on deep learning

Abstract

The invention discloses a lumbar intervertebral disc MRI image classification method based on deep learning, and belongs to the technical field of medical image classification. According to the invention, an edge enhancement module based on fast Fourier transform is constructed, edge feature extraction is carried out by using a Sobel operator, Fourier transform is carried out to a frequency domain, the frequency domain is decomposed into low, medium and high frequency bands, and multispectral matching dynamic weight adjustment fusion is used to improve the target edge capture capability; a fine-grained parallel feature extraction module is provided, through multi-path parallel convolution and feature fusion, the overall shape change of a lumbar intervertebral disc region is captured by using a large-scale convolution kernel, and local details and texture features are extracted by using a small-scale convolution kernel, so that the model can extract features of different scales more sufficiently; a self-adaptive graph convolution module is designed, the relation between lumbar vertebra MRI image feature channels is modeled with the help of a graph convolution network, channel relevance is learned in a self-adaptive mode, and the capacity of capturing the mutual relation between the channels is higher; according to the method, the precision and robustness of the lumbar disc degenerative change in identification are effectively improved.

Images