Diabetic Eye Disease Classification by Residual Network based Feature Mapping with Support Vector Machine

Preview this chapter:

Diabetic Eye Disease Classification by Residual Network based Feature Mapping with Support Vector Machine, Page 1 of 1

< Previous page | Next page > /docserver/preview/fulltext/9789815324099/chapter-4-1.gif

The retinal blood vessels' diameter and tortuosity will alter as a result of diabetic retinopathy. The prediction of differences in retinal blood vessel diameter and new vessel formation is the desired focus of investigation. Segmenting the retinal blood vessels is necessary in order first to observe the alterations. The suggested system improves the quality of the segmentation results over diseased retinal images. A generative and non-generative deep learning model is proposed in this study. The CNN-SVM was separately applied in the experiment. For classification tasks, the CNN-GMM-SVM model that has been suggested does have a sensitivity of 81.0%. When compared to other models, the CNN-GMM-SVM model that has been suggested produces the best outcomes. The CNN-GMM-SVM model increases classification sensitivity by 5.4% when compared to CNN-SVM and CNN-GMM.