Rapid and Accurate Abnormal Oct Image Detection Using Ml-Based on Intensity Line Profile Features

Optical Coherence Tomography (OCT) is a non-invasive medical imaging technique that uses light waves to capture high-resolution cross-sectional images of the retina. While most existing OCT abnormal detections rely on complex deep-learning approaches, this paper aims to accelerate the process by utilizing intensity line profiles to identify abnormal OCT images. An intensity line is a set of intensity values derived from vertical stretches of each pixel. This method offers a fast, less resource-intensive, and straightforward approach to automatically detecting abnormal OCT images. This study explores features from intensity line profiles and employs four classic classifiers: decision tree, KNN, SVM, and AdaBoost, to differentiate between normal and abnormal OCT images. The SVM achieves an accuracy of 90.5 % with a false negative rate of 9.9 % which showed the most proficient performance in identifying abnormalities, highlighting their applicability in OCT diagnostics. In terms of healthcare benefits, our method reduces costs and time for OCT image analysis, making OCT diagnostics more affordable and decreasing the workload of medical staff.