GazeDx: Interactive Visual Analytics Framework for Comparative Gaze Analysis with Volumetric Medical Images

Area of Science:

• Medical Imaging Analysis
• Human-Computer Interaction
• Visual Analytics

Background:

• Comparative analysis of gaze data is crucial for understanding radiologist image interpretation.
• Existing methods for gaze analysis in volumetric imaging lack interactive tools.
• 3D perception in volumetric images increases cognitive load, challenging gaze pattern comparison.

Purpose of the Study:

• To introduce GazeDx, an interactive visual analytics framework for comparative gaze analysis of volumetric images.
• To integrate contextual information (pupil size, windowing) with gaze data for deeper insights.
• To address the limitations of static image comparisons in volumetric gaze analysis.

Main Methods:

• Developed GazeDx, an interactive visual analytics framework.
• Integrated contextual information like pupil size and windowing into gaze data analysis.
• Designed a context-embedded interactive scatterplot for diverse gaze data examination.

Main Results:

• GazeDx facilitates interactive comparison of gaze patterns from multiple readers on volumetric images.
• The framework successfully integrates contextual data, enabling more in-depth analysis.
• Case studies with experienced radiologists demonstrated the utility of GazeDx for comparative gaze analysis.

Conclusions:

• GazeDx provides an effective solution for comparative gaze analysis in volumetric medical imaging.
• The integration of contextual information enhances the ecological validity of gaze data findings.
• Interactive visualization tools are essential for advancing the understanding of radiologist visual behavior.