Noval Approach for Retinal Disease Detection System
Aditya Anand*1, Satya Prakash*2, Ashish Kumar Srivastava*3
1234Department of Information Technology,
1234IIMT College Of Engineering, Greater Noida, Uttar Pradesh, India.
adityaanand.it@gmail.com
sp5495302@gmail.com
hiashish2006@gmail.com
ABSTRACT
This research paper explores the complex terrain of developing an automatic retinal disease detection system in an attempt to break down the core characteristics, technological paradigms, and challenges involved in emulating a powerful AI-driven diagnostic tool. The system detects diabetic retinopathy and glaucoma, two major causes of blindness, through the utilization of deep learning and computer vision methodologies. The main goals of this research include an in-depth examination of the system's major features, investigation of appropriate technologies for medical image classification, and an in-depth examination of challenges encountered while developing such a system. The technological setup includes front-end development through Streamlit for user interface, with back-end based on Python with TensorFlow/Keras for model implementation. Real-time image examination is enabled with CNN-based structures, and model outputs and storage of data are handled using cloud and local storage solutions. Authorization and authentication are integrated where relevant to provide medical data secure processing. The system development challenges lay in data quality, model correctness, interpretability, and incorporating it into the clinical workflow. Overcoming these challenges successfully is essential for developers who want to develop a scalable, secure, and reliable medical diagnostic device. This study empowers developers and healthcare technologists with an in-depth guide, providing insights into design principles, technology selection, and how to overcome challenges in the ever-evolving space of AI-based retinal disease detection.
KEYWORDS
Retinal Disease Detection, Diabetic Retinopathy, Glaucoma, Convolutional Neural Network (CNN), MobileNetV2, EfficientNetB0, Deep Learning, Fundus Images, Feature Extraction, Medical Image Classification, Transfer Learning, Automated Diagnosis
