Intelligent Papilledema Detector (IPD)

Background: Timely diagnosis of papilledema is essential to avoid vision loss and progress of life threatening conditions. Expert ophthalmologists or neurologists are not available in Emergency departments and in rural healthcare centers for timely detection. An intelligent, non-invasive detection system to aid healthcare professionals to detect papilledema and triage neurological patients is essential for early diagnosis for saving vision & even livesMethodology: Retinal fundus images are used to identify papilledema. After suitable preprocessing of the data, trained Convolutional Neural Networks can be used to classify the images to detect papilledema. Our proposed model uses EfficientNet-B3 to accurately and efficiently detect papilledema using an image dataset.Results: Accuracy of 98.54% is achieved with the EfficientNet-B3 model. Other performance metrics are also significantly higher than existing literature.Conclusion: The Intelligent Papilledema Detector will be very helpful in emergency departments and rural healthcare centers to aid with early detection of papilledema. The results obtained are very encouraging, though training with more data from various sources will help improve the practical usability of the system. Emerging trends of using smartphones with a lens assembly to capture also can be taken up as further work.