Medical image grading system and method based on deep convolutional neural network

Abstract

The invention discloses a medical image grading method and system based on a deep convolutional neural network. The method comprises the steps that an original fundus image is automatically partitioned into an optic disk image and an optic cup image; image green channel components are extracted; a histogram is utilized to correct extracted grayscale images in a balanced mode; the two features of acup-to-disk ratio and optic nerve fiber layer defects are extracted respectively; a deep convolutional neural network algorithm is used to train multiple sub-classifiers; and the sub-classifiers arecombined, and a final classification result is obtained through voting. By the adoption of the technical scheme, classification accuracy is remarkably improved, it is beneficial for reducing misdiagnosis, and therefore the practical value of the classifiers is raised.

Description

technical field [0001] The invention belongs to the field of machine learning, and in particular relates to a grading system and method for medical images (for glaucoma fundus images) based on deep convolutional neural networks Background technique [0002] Glaucoma is the second leading cause of blindness in the world. Patients with glaucoma often have large optic disc ratios, choroidal atrophic arcs, and nerve fiber layer defects. Glaucoma screening is a very complex and difficult task. At present, the diagnosis of glaucoma basically relies on manual observation. But the human visual system has its deficiencies, such as subjectivity, limitation, fuzziness, lack of persistence and so on. In order to realize intelligent automation and informatization of detection, a computer image recognition and diagnosis system that can simulate human visual function and exceed its performance is urgently required to identify and diagnose glaucoma lesions. Contents of the invention ...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the problem of glaucoma fundus image recognition, the present invention provides a medical image classification system and method based on deep convolutional neural network

Images