Focus improves diagnosis of brain tissue during surgery

FOCUS: A Subspecialty Foundation Model for Intraoperative Central Nervous System Pathology

Computer Vision and Pattern Recognition

Summary

Diagnosing brain tissue samples during surgery is very hard because the samples may have artefacts and decisions must be fast. The authors created FOCUS, a specialized AI model trained on many frozen brain tissue samples to help with these diagnoses. It performs very well on many diagnostic tasks, including detecting lymphoma, which is important during surgery. When junior doctors used FOCUS, their accuracy improved to match or beat expert doctors. This tool could help surgeons make better decisions during brain operations.

What this means in practice

  • For intraoperative neuropathologists: Assist in diagnosing CNS frozen tissue during surgery to improve accuracy and sensitivity, especially for lymphoma detection.
  • For clinical pathology labs: Integrate a specialized AI tool to provide provisional molecular information and detailed diagnoses to inform surgical decisions.
  • For medical device companies: Develop commercial intraoperative pathology diagnostic software incorporating FOCUS to support CNS tissue analysis.$Commercial implications: FOCUS enables robust AI-powered CNS pathology assistance, making advanced diagnostic tools marketable to hospitals and surgical centers.

Authors

Xing Liu, Yifan Guo, Hongtai Lv, Lei Lou, Yuyi Tan, Jiang Du, Xijia Zhang, Ming Zhang, Chao Pang, Minglei Li, Wei Liu, Xiaobo Zhu, Song Zhang, Rong Yu, Hanlin Shu, Wenzhuo Xu, Tao Jiang, Cheng Lu, Jie Tian, Qing Chang, Zhenyu Liu

Abstract

Intraoperative frozen-section diagnosis of central nervous system (CNS) lesions guides surgical decisions but is challenged by artefacts, diagnostic complexity and time constraints. Most pathology foundation models have limited exposure to CNS frozen tissue. Here we present FOCUS, a subspecialty foundation model pretrained on 27,930 frozen-sections enriched for CNS tissue. FOCUS led 11 internal diagnostic and 12 biomarker-prediction tasks, with mean AUCs of 0.969 and 0.877, respectively, retaining leading performance in temporal and external cohorts. To provide more comprehensive diagnostic information, we further organized ten diagnostic tasks into a WHO-aligned diagnostic hierarchy and separately evaluated FOCUS for lymphoma detection, a critical task for intraoperative surgical decision-making. In reader assessments by six neuropathologists, FOCUS assistance enabled junior readers to approach or exceed unassisted senior readers in hierarchical diagnosis and raised junior readers' lymphoma sensitivity from 81.6% to 94.7%. FOCUS could improve diagnostic accuracy and granularity, inform lymphoma-related resection decisions and provide provisional molecular information. Such capabilities highlight the potential of FOCUS to provide robust, clinically actionable support for intraoperative CNS pathology.