Topimage rigidle
In introduction to HCI class.

Comparative Gaze Analysis Framework for Volumetric Medical Images

Hyunjoo Song, Jeongjin Lee, Tae Jung Kim, Kyoung Ho Lee, Bohyoung Kim, and Jinwook Seo / 2014


Project Description

We present an interactive visual comparison framework (GazeDx) for gaze data from multiple readers, which incorporates important contextual information into the comparative analysis process. A comparative analysis of gaze pattern is essential to understand how radiologists read medical images. However, most prior work on volumetric medical images focused on visualization of gaze patterns, but did not address the need for comparative analyses of multiple readers’ gaze patterns. The GazeDx framework supports qualitative comparison based on interactively coordinated multiple views (spatial view with 3D gaze visualization, enhanced navigation charts, and matrix view), and quantitative comparison of gaze patterns in the similarity view with several similarity measures. It also integrates crucial contextual information such as pupil size, distance to a monitor, or windowing (i.e. adjustment of image contrast and brightness which affects visibility of organs and lesions) into the analysis process.