AI-Pathologist Framework Improves Accuracy and Reliability in Cancer Diagnosis

Ensuring reliable cancer diagnosis from digital pathology remains a critical challenge as clinical decisions rely on accurate slide interpretation. While artificial intelligence (AI) has accelerated whole-slide image review, many models lack mechanisms to verify the dependability of their outputs in high‑stakes settings. Pathologists also need tools that can recognize uncertain cases and support defensible triage. 

At The Hong Kong Polytechnic University (PolyU), a research team has developed TRUECAM (TRustworthiness-focused, Uncertainty-aware, End-to-end CAncer diagnosis with Model-agnostic capabilities), an integrated framework for whole-slide image analysis in pathology. Designed for non-small cell lung cancer (NSCLC) subtyping, the framework is model‑agnostic and spans the pipeline from slide input to diagnostic output. It assesses confidence in predictions, flags out‑of‑scope inputs, and prompts review when uncertainty is high, enabling a structured collaboration between AI systems and pathologists.

TRUECAM incorporates multiple trustworthiness safeguards. It automatically removes ambiguous or uninformative regions in whole-slide images to preserve diagnostically relevant content. It also applies conformal prediction to keep diagnostic error rates within an acceptable range. Whole-slide imaging (WSI), which digitally scans glass slides into high‑resolution images for “virtual microscopy,” is the primary context for the framework’s operation.

In systematic evaluations across multiple cancer datasets, the team tested TRUECAM with both task‑specific models and larger foundation models. Computational experiments showed that models wrapped with TRUECAM consistently outperformed their unwrapped counterparts in classification accuracy, robustness, interpretability, data efficiency, and fairness. Although developed for NSCLC subtyping, the approach was additionally applied to breast, brain, and kidney cancer subtyping, as well as a 46‑class pan‑cancer slide‑level classification task, indicating broad applicability described by the team.

The study is published in Nature Biomedical Engineering on June 23, 2026. According to the investigators, TRUECAM can be integrated into pathology AI models of various sizes, architectures, purposes, and complexities to support responsible diagnostic use in clinical environments.

“TRUECAM strikes a sound balance between fully pathology AI-powered and purely pathologists-led cancer diagnosis. When the model’s confidence in its diagnostic outputs is high, the system can help handle clear-cut cases, while uncertain cases are flagged and passed on to pathologists for further review and clinical judgement. This AI–pathologist collaboration helps improve diagnostic efficiency, lighten pathologists’ workloads, enhance diagnostic reliability and scale up diagnostic capacity,” said Prof. Zhang Xiaoge, Assistant Professor of the Department of Industrial and Systems Engineering at The Hong Kong Polytechnic University.

“While the published study focuses mainly on whole-slide images, we are further exploring additional modalities, such as molecular profiles including RNA-sequencing and diagnostic reports, to broaden the framework’s potential scope. TRUECAM provides a systematic solution to building trustworthy pathology AI and strengthens the foundation for deploying it in real-world settings,” added Prof. Zhang.

Related Links
PolyU