Core Concepts
Eye-tracking technology enhances medical image analysis by providing valuable insights into radiologists' visual attention patterns, improving diagnostic accuracy and interpretability of deep learning models.
Abstract
Eye-tracking technology offers significant promise in enhancing healthcare-related tasks, particularly in medical image analysis. It provides valuable insights into human visual attention patterns, transforming how healthcare professionals engage with and analyze diagnostic images. This systematic review explores the integration of eye-gaze data with machine learning and deep learning approaches to enhance diagnostic accuracy and efficiency in medical imaging. The study investigates various applications such as image classification, model interpretability enhancement, data annotation, pathology detection, segmentation, and assessing expertise fatigue and skill levels.
Key points include:
Eye-gaze tracking research enhances healthcare tasks by providing insights into human visual attention patterns.
Eye-tracking data improves understanding of radiologists' cognitive processes during diagnosis.
Various methodologies like fixation, saccades, scan paths are used to capture eye movement parameters for analyzing visual search patterns.
Eye-gaze tracking applications categorized into image classification, model interpretability enhancement, data annotation, pathology detection, segmentation.
Studies utilize conventional machine learning algorithms, convolutional neural networks (CNN), transformer-based methods for analyzing eye-gaze data in medical imaging.
Integration of eye-tracking data with ML/DL models enhances diagnostic accuracy and interpretability in medical imaging.
Stats
Teng et al. [31] utilized LightGBM, XGBoost, CatBoost for F-Scores.
Martinez et al. [22] employed AlexNet with Cross-entropy loss function for AUC and Accuracy metrics.
Stember et al. [30] used U-Net architecture with Binary cross-entropy and Dice coefficient losses for F-Score evaluation.
Quotes
"Eye-gaze tracking research extends beyond controlled environments into day-to-day scenarios." - Content
"Understanding eye movement parameters is crucial to capturing research matter." - Content
"Accurate visual data interpretation is significant for diagnostic decision-making in various medical fields." - Content