Lung nodule detection using eye-tracking

This paper describes a decision support system for determining salient features for CT lung nodule detection using an eye-tracking based machine learning technique. The method first analyses the scan paths of expert radiologists during normal examination. The underlying features are then used to highlight salient regions that may be of diagnostic relevance by merging visual features learned from different experts with a weighted probability function. The framework has been evaluated using data from CT lung nodule examination and the results demonstrate the potential clinical value of the proposed technique, which can also be generalized to other diagnostic applications.