A Transfer Learning Based Detection and Grading of Cataract using Fundus Images

Among the most prominent causes for visual impairment, especially among older adults, is cataract. As per the World Health Organisation (WHO), there are 2.2 billion individuals globally are estimated to be blind or possess an eyesight problem. One of the most identified and important causes of this is cataracts. Cataracts should be identified and treated as soon as possible to avoid blindness. Ophthalmologists use an expensive slit lamp to diagnose cataracts in regions with few medical facilities. Consequently, the issue is that a lack of skilled ophthalmologists may delay the identification of cataracts, for which medical treatment is unavoidable. Medical image analysis based on artificial intelligence provides a rapid and precise diagnosis in modern healthcare. We utilized deep learning models based on transfer learning, namely VGG19, and ResNet-50, to diagnose cataracts using fundus images and enhance classification accuracy. The performance metric for the model is accuracy; the highest achieved accuracy was 98%.