Eye tracking applications in computer aided diagnosis and image processing in radiology
Inventors
Wood, Bradford J. • Celik, Haydar • Bagci, Ulas • Turkbey, Ismail Baris
Assignees
US Department of Health and Human Services • University of Central Florida Research Foundation Inc
Publication Number
US-10839520-B2
Publication Date
2020-11-17
Expiration Date
2038-03-05
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
A system and method for using gaze information to extract visual attention information combined with computer derived local saliency information from medical images to (1) infer object and background cues from a region of interest indicated by the eye-tracking and (2) perform a medical image segmentation process. Moreover, an embodiment is configured to notify a medical professional of overlooked regions on medical images and/or train the medical professional to review regions that he/she often overlooks.
Core Innovation
This disclosure presents a method and system that employ gaze information from eye-tracking technology combined with computer derived local saliency and gradient information to perform real-time medical image segmentation. The inventive process capitalizes on radiologists' visual attention captured by eye-tracking to directly define object regions of interest in radiology scans, overcoming manual or automated recognition challenges traditionally associated with segmentation tasks.
The problem addressed arises from difficulties in analyzing volumetric medical images, such as lung CT scans, where prior eye-tracking approaches struggled with volumetric synchronization and the complexity of extensive eye-tracking data. Conventional segmentation demands either manual interaction via mouse or exhaustive automated recognition, both time-consuming and inefficient. This disclosure resolves these issues by mapping eye gaze to identify attention regions, integrating saliency and gradient data to infer foreground and background cues, culminating in effective and efficient image segmentation.
Claims Coverage
The patent discloses two independent claims covering a method and a system for automatic medical image segmentation using eye-tracking technology. The claims focus on features integrating gaze data with saliency and gradient information to identify and segment regions of interest.
Integration of eye-tracking data with medical image segmentation
A method/system using an eye-tracker with an eye-facing camera and scene-facing camera to record gaze information and transfer it to a workstation for processing.
Creation of two-dimensional visual attention maps from gaze information
Constructing visual attention maps from gaze data to identify regions of the medical images where the user’s attention meets predefined criteria.
Identification of foreground and background cues using local saliency and gradient information
Obtaining local saliency and gradient information from medical images corresponding to attentive regions, then identifying foreground and background cues based on this information for each region.
Automatic segmentation of objects of interest based on identified cues
Employing the background and foreground cues to perform automatic segmentation of objects associated with the regions of interest.
The invention claims combine real-time eye-tracking data with image processing techniques to enable automatic identification and segmentation of medical image regions, enhancing segmentation processes through attention-driven foreground and background cue detection.
Stated Advantages
Provides improved accuracy and efficiency in medical image segmentation by using eye-tracking data to guide recognition of regions of interest.
Enables real-time quantification and analysis of radiology scans without manual or exhaustive automated recognition steps.
Supports identification of overlooked regions by medical professionals and facilitates personalized training to reduce diagnostic misses.
Documented Applications
Real-time image segmentation in radiology scans including X-rays, MRI, PET, ultrasounds, and CT images.
Interventional radiology and surgery room applications requiring immediate image segmentation.
Training tools for radiology residents to improve diagnostic accuracy by highlighting frequently overlooked regions.
Personalized models to identify and learn radiologists' weak spots or differences in interpretation compared to peers.
Interested in licensing this patent?