Medical image identification method and device

Abstract

The invention provides a medical image identification method and device. The medical image identification method includes the steps: acquiring a medical image; classifying the medical image by using afirst machine learning model to obtain a first vector, wherein the first vector indicates a first confidence that the type of the medical image is healthy or abnormal; utilizing a second machine learning model to classify the medical image to obtain a second vector, wherein the second vector indicates a second confidence that the type of the medical image is various disease types; according to the first vector and the second vector, obtaining a third vector; and obtaining an identification result of the medical image according to the third vector.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a medical image recognition method and equipment. Background technique [0002] Using machine learning algorithms and models to recognize images is an efficient way, and it is also the underlying technology in many fields such as autonomous driving, smart cameras, and robots. [0003] Medical images can usually reflect various types of diseases. For example, fundus images can reflect various eye diseases such as hemangioma, fundus hemorrhage, and glaucoma. Using machine learning models (such as neural networks) to identify medical images requires first using sample images to train the model, which requires a large number of annotated pictures and relies heavily on annotators. Different countries, cities, and doctors have different definition standards for some diseases, and the level of annotation personnel is uneven. The quality of the annotated images is relatively larg...

AI Technical Summary

Problems solved by technology

Different countries, cities, and doctors have different definition standards for some diseases, the level of labeling personnel is uneven, and the quality of the labeled images varies greatly. Wrong and inaccurate labeling content will cause data noise

[0004] The existing artificial intelligence-based medical image recognition schemes use the models trained by these labeled images to obtain recognition results, which are greatly affected by data noise, and the accuracy of the recognition results is not high enough, which may even cause the failure of model training.

Images