Machine Learning-Based Diabetic Retinopathy Detection: A Comprehensive Study Using InceptionV3 Model
Gautam A. Deshpande, Yash Govardhan, Anamika Jain
Abstract
Diabetic retinopathy is considered as a common eye disease that affects vision of people those have diabetes. Early diagnosis of diabetic retinopathy has become a crucial step to prevent vision loss. In this paper, we have proposed a method to detect diabetic retinopathy using machine learning based method. We used two publicly available datasets, EyePACS and APTOS 2019, for training and testing our model. We employed a pre-trained model, Inception V3,and fine-tuned it on our dataset. We achieved and accuracy and F1 score of 74.28% and 73.81 % respectively on the EyePACS dataset and on APTOS 19 dataset we obtained an accuracy and F1 score of 81.61 % and 80.21 % respectively. Our findings suggest that with machine learning, we can detect diabetic retinopathy in the early stages.