System and method for improving image recognition precision on premise of privacy protection

Abstract

The system for improving the image recognition precision on the premise of privacy protection comprises a plurality of user node modules and a center node module, according to the method for improving the image recognition precision on the premise of privacy protection, by constructing a twin noise image library of a local medical image library of each user node module, the purposes of protecting the privacy of a patient and concentrating dominant medical image training data of each hospital are achieved, so that a medical image recognition model with a high training level is obtained, and the medical image recognition precision is improved. And the automatic identification accuracy of the medical image is improved.

Description

technical field [0001] The invention relates to a system and method for improving image recognition accuracy under the premise of privacy protection, belonging to the field of information technology, in particular to the technical field of medical image recognition. Background technique [0002] With the extensive and long-term development of deep learning technology, more and more computer tasks are solved or partially solved by deep learning, including tasks related to segmentation and recognition of large-scale medical images. People hope to use the current deep learning neural network technology to help hospitals realize automatic segmentation and automatic identification of medical images, thereby improving the processing speed of medical images and allowing patients to get faster and more accurate diagnosis. [0003] At present, model training based on deep learning requires a large amount of high-quality data to support its performance. Specifically, training a model...

AI Technical Summary

Problems solved by technology

[0005] However, due to the need for patient privacy protection, it is currently difficult to gather the medical image training data of various hospitals and conduct unified training for the medical image recognition model.

The result will lead to: if only one hospital's medical image training data is used to train the medical image recognition model, then the final medical image recognition model will only have better performance on data similar to the hospital's data set, and other In the data scenarios of different parts of the body (such as CT and MRI images of different parts of the body), the performance may be very poor. Even in the case of insufficient data in the hospital, it is difficult for CT, MRI images of any body parts, and medical image recognition models. good recognition effect

Images