Researchers have introduced PathoGaze1.0, a new dataset designed to improve AI's capabilities in pathology diagnostics. This dataset captures detailed eye-tracking and navigation data from 11 pathologists as they interpreted 397 whole-slide images from the CAMELYON16 dataset. The collected data includes over 18 hours of eye-tracking, navigational scanning, and 1,140 diagnostic decisions, all recorded using a specialized testbed called PTAH. The goal is to provide behavioral insights into diagnostic errors and inconsistencies, which are common in current pathology practices. AI
IMPACT This dataset could enable AI models to better understand and replicate human diagnostic processes in pathology, potentially improving accuracy and consistency in cancer detection.
RANK_REASON The item describes a new dataset released via arXiv, which is a form of research output. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →