Method for identifying medical image, method for model training and server

Abstract

The embodiment of the invention discloses a method for identifying a medical image. The method comprises the steps that a to-be-identified medical image set is obtained, wherein the to-be-identified medical image set comprises at least one to-be-identified medical image; a to-be-identified area corresponding to each one to-be-identified medical image in the to-be-identified medical image set is extracted, wherein the to-be-identified area belongs to a part of the to-be-identified medical images; the recognition result of each to-be-identified area is determined through a medical image recognition model, the medical image recognition model is obtained according to training of a medical image sample set, the medical image sample set comprises at least one medical image sample, each medical image sample carries corresponding annotation information, the annotation information is used for indicating the types of the medical image samples, and the recognition result is used for indicating the types of the to-be-identified medical images. The embodiment of the invention also discloses a method for model training and a server. The method for identifying the medical image, the method for model training and the server greatly save the manual annotation cost and time cost, and have stronger reliability and credibility.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a medical image recognition method, a model training method and a server. Background technique [0002] With the emergence and rapid development of medical imaging technologies such as computed tomography (CT), magnetic resonance imaging (MRI) and ultrasound (ultrasonic, US), a large number of clinical diagnostic data are generated and stored in hospitals. and analyzed medical images. In recent years, with the rapid development of computers and related technologies and the maturation of graphics and image technology, medical workers can observe medical images from multiple directions, levels, and angles, thereby assisting doctors to diagnose lesions and other sensitive images. Focused analysis on the region of interest improves the accuracy of clinical diagnosis. [0003] At present, for the identification of medical images, it is mainly through medical workers to mark me...

AI Technical Summary

Problems solved by technology

[0004] However, a large number of medical images also need to consume a lot of time for medical workers to label, resulting in high labeling costs

In addition, manual labeling of medical images is prone to errors, which reduces the reliability and accuracy of medical image recognition

Images