A Machine Learning Framework for Classification of Expert and Non-Experts Radiologists using Eye Gaze Data

Artificial intelligence (AI) techniques enhance the efficiency of radiology data interpretation and reporting process. The paper focuses on classifying a participant as an expert or non-expert based on the features related to saccade, such as saccade latency, saccade amplitude, etc. These features help to understand whether the participant knows the respective field or not. The data used here is how the participant looks at an optic disc image and whether they know where to look precisely. The proposed work used different machine learning models to classify experts and non-expert radiologists and observed that the decision tree gave the highest accuracy of 96%. Finally, we create a dashboard with the help of powerbi, which displays the insights about the expert and non-expert viewing the optic discs.