DISEASE CLASSIFICATION OF MACULA OCT SCANS USING CNN
Abstract
The timely and precise identification of macula diseases plays a vital role in effective treatment and the prevention of vision impairment. Optical Coherence Tomography (OCT) imaging has become a cornerstone in diagnosing various macula diseases. However, the manual interpretation of OCT images by ophthalmologists is not only time-intensive but also susceptible to human error. To address this challenge, a Convolutional Neural Network (CNN) architecture is meticulously crafted and trained using a meticulously annotated dataset. The model undergoes optimization employing appropriate loss functions and optimization algorithms, coupled with extensive experimentation and hyper parameter finetuning to attain optimal performance. Subsequently, the efficacy of the trained model is assessed on an independent test set of OCT images, and its performance is juxtaposed with that of expert ophthalmologists. Preliminary findings highlight the promising potential of the proposed deep learning- based system in accurately categorizing macula diseases from OCT images. Such a system holds promise in aiding ophthalmologists to render more precise and prompt diagnoses, facilitating early interventions and ultimately enhancing patient outcomes. Moreover, the system presents a potential solution to the scarcity of expert ophthalmologists in certain regions by serving as a cost-effective screening tool for macula diseases
Keywords
Macula diseases, Optical Coherence Tomography, Deep learning, Convolutional Neural Networks, Disease classification, Ophthalmology