Visual Search Patterns in 3D Pancreatic Imaging: An Eye Tracking Study

2025-06-23 · Source: cs.CV updates on arXiv.org · Field: Science & Research — Health & Medical Research, Research Methodology & Innovation · Depth: Expert, long

Summary

This study investigates visual search patterns and strategies employed by radiologists when assessing the pancreas in 3D computed tomography (CT) images, addressing a gap in eye tracking research for volumetric medical imaging. A pilot study involved two radiologists reviewing both healthy and cancerous pancreatic CT images using Pupil Labs Neon eye-tracking glasses, with gaze data aligned to on-screen content and pancreas segmentation performed by TotalSegmentator. The analysis revealed distinct gaze behaviors and organ coverage differences between the radiologists; one predominantly used a "scanning" approach, while the other adopted a unique strategy not directly classifiable as "scanning" or "drilling." Observed variability included differences in mean visit durations, median visit times, and the number of switches to the pancreas region, indicating individual search strategies. Due to the limited sample size, the study concludes that larger investigations are necessary to robustly confirm and further explore these preliminary findings.

Key takeaway

A pilot eye-tracking study on 3D pancreatic CT interpretation identified distinct visual search strategies among radiologists. One radiologist employed a "scanning-oriented" approach with more frequent, shorter visits (e.g., mean 58 visits, median 0.96s with peripheral gaze), while another used a "one-organ-at-a-time" targeted strategy with fewer, longer visits (e.g., mean 28 visits, median 1.39s). These preliminary insights highlight the need for larger studies to characterize diverse volumetric search patterns, which could inform training and optimize diagnostic efficiency in radiology.

Topics

• Eye Tracking
• 3D Medical Imaging
• Pancreatic Imaging
• Visual Search Strategies
• CT Scans

Code references

• pupil-labs/dynamic-rim-module

Best for: AI Scientist, Research Scientist, Domain Expert

Related on AIssential

• When is 3D Worth It? A Resource-Performance Frontier for CNNs and Transformers in Lung CT — Input dimensionality significantly impacts model performance and stability in volumetric medical imaging.
• Where are they looking in the operating room? — Gaze-following in the OR significantly enhances surgical workflow analysis and team communication understanding.
• RARE disease detection from Capsule Endoscopic Videos based on Vision Transformers — Vision Transformers can be fine-tuned for multi-label classification of gastrointestinal diseases from capsule endoscopic videos.
• CrossPan: A Comprehensive Benchmark for Cross-Sequence Pancreas MRI Segmentation and Generalization — Cross-sequence domain shift is the primary barrier to robust pancreas MRI segmentation.
• A 3D SAM-Based Progressive Prompting Framework for Multi-Task Segmentation of Radiotherapy-induced Normal Tissue Injuries in Limited-Data Settings — A 3D SAM-based framework improves radiotherapy injury segmentation using progressive prompting and a small-target focus loss.
• Pan-FM: A Pan-Organ Foundation Model with Saliency-Guided Masking for Missing Robustness — Pan-FM is a pan-organ foundation model robust to missing data via saliency-guided masking.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.