The advent of scientific research and technologies in the domain of medicine has recently taken limelight due to its immense benefit on humankind and the medical community. The expert faculty and scholars have recently published a patent, “A System and a Method for Early and Accurate Diagnosis of Cataracts” with Application no.: 202341072058 has put forth an engaging invention on utilising technology for early cataract recognition. Hearty congratulations to Prof. Siva Sankar Yellampalli, Professor, Dr Ramesh Vaddi, Associate Professor, and their Ph.D. Scholars Ms P L Lahari, Mr P Rahul Gowtham, and Mr A Vinod Kumar from the Department of Electronics and Communication Engineering for this groundbreaking achievement!
Abstract
Cataracts are a common eye condition in which the lens of the eye gets clouded, impairing vision. Early cataract detection is crucial for prompt treatment and vision preservation. We have classification and prediction algorithms like VGG, ResNet, DenseNet, Xception, Inception, and other object detection techniques like Yolo, Fast R-CNN, and SSD. Various pre-trained models are employed for cataract categorisation and prediction. Several attempts to detect cataracts have been made, but none have proven effective. A clinical examination by eye specialists is used to diagnose cataracts. An edge board can be used instead of a clinical examination to diagnose cataracts.
We created a method for real-time cataract recognition using the present pre-trained weights of the object detection model YoLoV5. We employ pre-trained YoLo V5 weights for model training, testing, and validation. Connect the Jetson Nano board and Lenovo HD USB camera to the CPU, which serves as the CPU. The monitor is used for programming, and the output is presented on the monitor owing to the board communicating with the camera. The result shows the image with an eye labelling box that tells if the eye is normal or cataract.