DIABETIC RETINOPATHY DIAGNOSIS USING KERNEL FUZZY C MEANS WITH CONVOLUTIONAL NEURAL NETWORK BASED RESIDUAL MODEL
Presently, Internet of Things (IoT) becomes popular owing to diverse its application scenarios like transports, building, healthcare, etc. This study introduces an efficient IoT based diabetic retinopathy (DR) diagnosis model using Kernel Fuzzy C Means Segmentation and Residual Network. The proposed model involves a sequence of processes namely image acquisition, pre-processing, segmentation, feature extraction and classification. At the initial stage, retinal fundus image acquisition takes place which captures the retina image of the patient using head mounted camera. Next, kernel fuzzy c-means (KFCM) based segmentation process is applied to identify the diseased area. Then, the features are extracted using convolutional neural network (CNN) based residual network (ResNet) model. Finally, softmax function is employed to carry out the classification task. The validation of the presented model takes place using Kaggle DR dataset and the experimental results verified the superior performance of the presented model. The obtained results indicated that the KFCM-CNNR model has resulted to a maximum accuracy of 96.89%, sensitivity of 93.12% and specificity of 98.16%.