NIHR BRC researcher Dr Tong Xin co-authored a new review paper drawing on expertise from the BRC-funded DeepCPath lab. The authors provide practical guidance for researchers developing the next generation of pathology AI tools.

Published in Frontiers in Digital Health, the paper, Translational AI in whole-slide image cancer histopathology: state of the art and regulatory-approved solutions, provides a comprehensive overview of the rapidly evolving field and highlights the gap between academic innovation and real-world clinical deployment.

While artificial intelligence is advancing, there are currently only four FDA-approved whole-slide image cancer solutions available for digital pathology. Progression into real clinical pathways remains blocked by pre-analytical tissue variations across scanners, false-positive alert burdens, and rigid validation frameworks for high-autonomy generative systems.

Crucially, this review exposes how standard academic metrics like AUC, a foundational evaluation metric in machine learning and statistics, can mask real-world clinical failures by hiding generalization losses on external datasets.

Dr Tong Xin said: ‘A significant gap still separates the emerging AI solutions from production-ready platforms in modern computational pathology. Bridging this gap requires aligning the distinct expectations of both fields. By deepening the collaborations between computing scientists and pathologists, wea are constructing this bridge on the stable foundations of a truly interdisciplinary ecosystem.’

The team at DeepCPath Newcastle aims to guide future developers through the complexities of regulatory compliance, validation methodologies, statistical benchmarks, and emerging state-of-the-art architectures to eventually bridge the translational gap between academic research and clinical deployment in digital pathology.

Based in the School of Computing at Newcastle University and embedded within the NIHR Biomedical Research Centre: Newcastle, the team brings together computer scientists, clinicians and domain specialists to address challenges in computational pathology, precision diagnostics and biomonitoring. Our work spans methodological research, translational partnerships with healthcare and industry, and training the next generation of interdisciplinary researchers.

This research was supported by the UKRI Newcastle Impact Accelerator Award 2024 and the NIHR Biomedical Research Centre (BRC): Newcastle, awarded to The Newcastle upon Tyne Hospitals NHS Foundation Trust, Faculty of Medical Sciences, Newcastle University, and Cumbria, Northumberland, Tyne and Wear Foundation Trust.

Read the paper