Deep learning based diabetic cornea image disease classification method

Abstract

The invention discloses a deep learning based diabetic cornea image disease classification method. The method comprises following steps: step A, downloading a data set of diabetic cornea images from akaggle website; step B, subjecting obtained diabetic cornea image data set to following pretreatments: color balance adjustment, brightness adjustment, contrast adjustment, zooming, and cutting; stepC, on the basis of a VGG network model, establishing a new VGG-L network classification model; step D, training the established model by the preprocessed image data set; and step E, judging the classification accuracy of the trained model by using a test set of diabetic cornea images. The adopted classification method can divide the diabetic cornea images into five groups according to the diseaselevels and has the advantages of high classification accuracy, good robustness, and good extensiveness.

Description

technical field [0001] The invention is a diabetic retinal image classification method, which is applicable to the fields of machine learning, biomedicine and image processing. Background technique [0002] In recent years, with the advancement of science and technology and the continuous improvement of medical imaging acquisition equipment, as well as the continuous development of image processing, pattern recognition, machine learning, deep learning and other disciplines, the field of multidisciplinary medical image processing and analysis (Medical Image Processing and Analysis) have achieved fruitful results. Diabetes is a group of metabolic diseases characterized by high blood sugar, which can easily lead to lesions in various tissues such as eyes, heart, brain, blood vessels, etc. It is a kind of fundus lesions with specific changes. Diabetic retinopathy, the most important manifestation of diabetic microangiopathy, is a major cause of blindness. At present, the clini...

AI Technical Summary

Problems solved by technology

However, the number of diabetic patients to be screened is huge, and medical resources are scarce in some areas. The vast majority of diabetic retinopathy patients cannot receive timely diagnosis and treatment, thus missing the best treatment opportunity, eventually causing irreversible visual impairment and even consequences of blindness

The effect of manual diagnosis is extremely dependent on the clinician's experience in diagnosis and treatment. Misdiagnosis and missed diagnosis may be caused by the patient's failure to seek medical treatment in time, mild lesions, and insufficient doctor experience. In severe cases, the patient may miss the best treatment period and cause blindness.

Images