Interactive thyroid whole slide image diagnostic system using deep representation

Pingjun Chen, Xiaoshuang Shi, Yun Liang, Yuan Li, Lin Yang, Paul D. Gader

Abstract: Highlights•We develop an interactive thyroid whole slide image classification and retrieval system using deep representation.•With the participation of pathologists, the proposed system can focus on the suspicious regions in thyroid frozen sections, thus speeding up the analysis and improving the diagnostic performance.•We propose retrieval of the whole slide image to base on the region of interest (ROI), which is more fine-grained, precise, and reliable.•We introduce a win-win scheme for pathologists and the AI system, enable the diagnostic system to be more easily accepted by both pathologists and patients.