Classification and grading of orthopedic disease lesions based on deep residual network

Abstract

The invention discloses a method for classifying and grading orthopedic disease lesions based on a deep residual network, which is specifically implemented according to the following steps: Step 1, offline learning, preprocessing professionally classified and marked orthopedic disease lesion images, and then performing Deep residual neural network training; step 2, online learning, back up the deep residual neural network trained in step 1 and deploy it to daily diagnosis and treatment, and use the training method of online learning to make the deep residual neural network pass daily The diagnosis and treatment data are constantly self-correcting. The method for classifying and grading orthopedic disease lesions based on the deep residual network of the present invention solves the problem that the accuracy of the existing convolutional neural network increases with the increase of the depth of the neural network structure, and then the accuracy decreases, and the offline learning method cannot be used for every day. There is a problem that self-correction cannot be performed as the number of consultations increases due to the use of medical treatment data.

Description

technical field [0001] The invention belongs to the technical field of classification methods for orthopedic disease lesions, and relates to a classification and grading method for orthopedic disease lesions based on a deep residual network. Background technique [0002] In the prior art, the classification and grading of lumbar intervertebral disc herniation, frozen shoulder, knee injury, and cervical spondylosis still rely on manual judgment on existing medical angiography. Now through CT, X-ray, nuclear magnetic resonance, and other medical imaging techniques, a large amount of medical diagnostic imaging data has been generated. But doctors personally cannot take full advantage. The current method still relies on the personal experience and ability of doctors to classify and grade these kinds of lesions by manual judgment. Therefore, it relies heavily on the personal experience and technical literacy of doctors, and the labor cost is extremely high. The accuracy of judgm...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method for classifying and grading orthopedic disease lesions based on a deep residual network, which solves the problem of the existing convolutional neural network with the increase of the depth of the neural network framework, the precision is saturated and then the precision drops, and the method of offline learning It is impossible to use the diagnosis and treatment data generated every day, so it cannot correct itself as the number of consultations increases

Images