A Medical Image Classification Method Based on Collaborative Deep Learning

Abstract

The invention discloses a medical image classification method based on collaborative deep learning, which is used to solve the technical problem of poor classification accuracy of the existing medical image classification method. The technical solution is to adopt a collaborative learning method between two deep convolutional neural networks, and to train in a paired learning mode. Each time the model accepts an image pair as input, and a pair of images is sent to the corresponding deep convolutional neural network. middle. These deep convolutional networks are initialized and trained by fine-tuning the pre-trained model, and a collaborative learning system is designed to enable two deep networks to learn mutually. The collaborative system is used to supervise the similarities and differences of image pairs, that is, whether they belong to the same category, and backpropagate the collaborative errors generated by the two deep convolutional networks in real time, and correct the weights of the networks, thereby further strengthening The ability of the network to learn feature representation can more effectively make accurate judgments on confusing samples.

Description

technical field [0001] The present invention relates to a medical image classification method, in particular to a medical image classification method based on collaborative deep learning. Background technique [0002] Medical image classification methods play an extremely important role in medical retrieval, literature review and medical research, and have always been a hot research issue in the field of computer-aided diagnosis and medical research. In the past decades of research by countless researchers, a complete set of image classification techniques under the traditional model has been formed. Its core elements are two parts: manual feature extraction and classifier design. Although there is a very complete theoretical system, it is difficult for traditional image classification methods to achieve the seamless combination of optimal features and optimal classifiers, which leads to a great impact on their performance. In recent years, the emergence of deep learning t...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the shortcomings of poor classification accuracy of existing medical image classification methods, the present invention provides a medical image classification method based on collaborative deep learning

Images