Diabetes retinopathy image multi-classification method based on deep learning

Abstract

The invention discloses a diabetes retinopathy image multi-classification method based on deep learning, and the method comprises the following steps: firstly, obtaining an original data set, and carrying out the preprocessing of the original data set; then carrying out category data balance on the preprocessed data set; then, carrying out migration training on the data set with balanced categorydata to obtain a diabetic retinopathy image multi-classification model; and finally, inputting the sample to be detected into the diabetic retinopathy image multi-classification model for prediction,and completing classification of the diabetic retinopathy image. The method provided by the invention has the advantage of being convenient to use, and in addition, the classification accuracy is improved by carrying out migration training on the data set after the class data is balanced.

Description

technical field [0001] The invention relates to an image classification method, in particular to a deep learning-based multi-classification method for diabetic retinopathy images, which belongs to the technical field of medical image recognition. Background technique [0002] In the current medical system, most of them tend to be manually detected, and the processing of diabetic retinal images is no exception. Diabetic retinopathy is divided into five types according to the severity level, namely non-diabetic retinopathy, mild diabetic retinopathy, Moderate diabetic retinopathy, severe diabetic retinopathy, and proliferative diabetic retinopathy. [0003] The traditional diabetic retina image processing method is to segment the system and use different processing techniques at different stages to achieve the goal. However, no matter which stage or which processing technique is used, the features are designed based on artificial experience. The specificity of a particular pr...

AI Technical Summary

Problems solved by technology

Although these features reflect the characteristics of the object to a certain extent, they only extract the underlying features, such as image edge features, grayscale features, etc. In addition, the structure of the retinal image itself is complex, and it is easy to cross-affect with various lesions. Influenced by complex background transformations, the method of processing retinal images is complicated and poor in generalization; in addition, this detection method relies on prior knowledge, especially experienced ophthalmologists

Therefore, the whole process not only consumes a lot of manpower, material resources and financial resources, but also is inefficient, and the diagnosis and treatment of retinopathy is greatly restricted, which eventually causes a large error in the image classification detection results and affects the multi-classification results.

Images