Classification method and device of medical image

Abstract

The embodiment of the invention discloses a classification method and device of a medical image. Firstly, division of training files and verification files is carried out, and then non-illness markervalues and illness marker values are added; in response to a training instruction on a classifier, the plurality of training files are used as input of an auto-encoder, convolution kernels used for classification training are generated through preprocessing the auto-encoder, and fine tuning is carried out on parameters in the classifier to obtain an optimal network structure according to the generated convolution kernels; and then, in response to a classification instruction of the classifier, the input to-be-classified medical image is classified according to the classifier obtaining the optimal network structure. An effective intelligent prediction model is established through a method of supervised learning, thus convolutional neural networks (CNN) can be applied to medical image classification, reference of assisted decision making is provided for medical staffs, and efficiency is improved while human resources are saved.

Description

technical field [0001] The invention relates to the field of medical detection, in particular to a method and device for classifying medical images. Background technique [0002] Medical imaging is widely used in clinical diagnosis and treatment. How to use a large number of medical images to assist doctors in the diagnosis and treatment of diseases is a problem that the industry is currently studying. An excellent medical image classification method must be based on the perfect and detailed classification of disease types and donors, so as to perform efficient retrieval, information analysis and mining at any time. Traditional medical images use manual recognition and text classification methods. However, with the increasing number of medical images, especially the differences in race, gender, and age involved, it is becoming more and more difficult for manual recognition. And the workload is increasing day by day. How to solve this problem, introduce increasingly mature ...

AI Technical Summary

Problems solved by technology

Each convolutional layer in the convolutional neural network is followed by a calculation layer for local averaging and secondary extraction. This unique feature extraction structure reduces the feature resolution.

Images