AI Pathology Analysis System Delivers Comprehensive Cancer Diagnosis

Accurate and timely cancer diagnosis is often limited by complex manual pathology workflows and lengthy turnaround times. Traditional pathology relies heavily on human interpretation, which can delay treatment and limit personalization for patients with complex cancers. To overcome these challenges, researchers have developed an advanced artificial intelligence (AI) system designed to streamline the entire pathology process—from diagnosis to personalized treatment planning.

The SmartPath system, built by researchers from The Hong Kong University of Science and Technology (HKUST, Hong Kong, China) in collaboration with leading clinical partners across Hong Kong and the Chinese Mainland, is undergoing multi-center clinical validation and represents a major milestone in AI-driven precision oncology. Trained on more than half a million whole-slide pathology images covering 34 tissue types, SmartPath is capable of assisting healthcare professionals with over 100 distinct clinical tasks.

Image: Prof. Chen Hao (right) and Prof. Liang Li (left) who led the research team that developed SmartPath (Photo courtesy of HKUST)

These include cancer classification, subtyping, biomarker quantification, treatment response evaluation, survival prediction, and the automated generation of detailed pathology reports. The system’s design is based on one of the largest and most diverse pathology datasets ever assembled, providing broad generalization across multiple cancer types such as lung, breast, colorectal, and gastric cancers.

At its core, SmartPath integrates two large-scale AI models that power its unique capabilities. The Generalizable Pathology Foundation Model (GPFM) provides a unified framework for tumor identification, subtyping, and biomarker quantification across cancer tissues, enabling personalized prognosis and treatment planning. The Multimodal Knowledge-enhanced Whole-slide Pathology Foundation Model (mSTAR) combines pathology imaging data with textual pathology reports and transcriptomic information.

This multimodal intelligence allows SmartPath to generate comprehensive pathology reports in minutes and even supports interactive visual question-answering for pathologists. Engineered as a fully integrated, end-to-end platform, SmartPath streamlines the cancer care cycle—from automated slide analysis and early risk alerts to AI-assisted reporting. By reducing diagnostic bottlenecks, the system allows pathologists to focus on high-complexity cases that require expert interpretation.

In comprehensive benchmarking and clinical validation, SmartPath achieved an accuracy of over 95% across multiple cancer types. The researchers continue to expand the system’s applications to rare and genetically complex malignancies, further enhancing its predictive capabilities and strengthening its impact on global precision medicine.

“The preliminary results from our recent trials are highly encouraging. SmartPath has demonstrated a remarkable ability to improve the accuracy of malignancy identification and provide reliable prognostic predictions,” said Prof. Liang Li, Director of the Department of Pathology at Nanfang Hospital, Southern Medical University. “The system's rapid generation of extensive preliminary reports has proven to cut down our diagnostic turnaround time significantly—a crucial factor in managing time-sensitive cancer cases. We are now witnessing the future of pathology—where AI-powered systems like SmartPath seamlessly integrate into the clinical workflow, augmenting our diagnostic precision and empowering pathologists to make more informed, data-driven decisions for precise and personalized patient care.”

Related Links:
HKUST