A method and device for training a classification network for medical images

Abstract

It is an object of the present invention to provide online training of classification networks for medical images. For a target disease, obtain an orthotopic chest image and its corresponding diagnosis report; locate the cardiopulmonary area in it according to the orthotopic chest image; The information is fed into the classification network as sample data to train it online. On the basis of using the medical image classification network trained offline, the present invention further obtains new medical images and diagnoses from the medical image database of the hospital by means of online learning in the actual scene where the medical image classification network is used. Report data for training, so that the efficiency of the classification network model can be further evolved and improved, and it is more suitable for its actual application scenarios and people, and by introducing the information of "onset location", the training accuracy is improved, and the performance of the classification network and online learning are significantly improved effectiveness.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to a technique for training a classification network used for medical images. Background technique [0002] Chest X-ray radiation is the most common method for the diagnosis of cardiothoracic diseases and is widely used clinically. In the prior art, the diagnosis by X-ray contrast still needs to rely on manual reading. Manual film reading has high requirements on the personal experience and ability of doctors; at the same time, manual film reading also has problems such as high cost, time-consuming, and easy to be interfered by human factors such as the doctor's status. [0003] With the rapid development of artificial intelligence, especially in the field of deep learning, a large number of researchers have tried to use this type of technology to help solve the diagnostic problems of medical imaging. When the neural network used for medical diagnosis obtained by ...

AI Technical Summary

Problems solved by technology

In the implementation process, the above schemes often only obtain the disease types corresponding to the images through simple keyword retrieval, but do not make full use of the key information of the location of the disease, resulting in poor performance of online learning

Images