China is digitizing pathology at scale. AI systems evaluated at Chinese cancer centers are matching specialist pathologist performance in several cancer types — with implications for diagnostic access.
Pathological diagnosis is the gold standard for cancer — tissue analysis by a trained pathologist determines diagnosis, subtype, and biomarker status that guides treatment. China has approximately 22,000 certified pathologists for a population of 1.4 billion — a ratio significantly lower than high-income country benchmarks. Combined with enormous cancer case volumes, this creates diagnostic pressure that AI is increasingly positioned to address.
The Pathology Capacity Gap — and the AI Response
Pathological diagnosis is the gold standard for cancer — tissue analysis by a trained pathologist determines diagnosis, subtype, and biomarker status that guides treatment. China has approximately 22,000 certified pathologists for a population of 1.4 billion — a ratio significantly lower than high-income country benchmarks. Combined with enormous cancer case volumes, this creates diagnostic pressure that AI is increasingly positioned to address.
China's digital pathology infrastructure — whole-slide image (WSI) scanners, laboratory information systems, and AI analysis platforms — has expanded rapidly since 2019. By 2024, over 600 hospitals had deployed digital pathology systems, creating the technical foundation for AI analysis at scale.
Gastric Cancer: AI Supporting Histological Classification
Gastric cancer is China's third most common cancer by incidence, and accurate histological classification — particularly Lauren classification and HER2 scoring — is critical for treatment selection. HER2-positive gastric cancer is eligible for trastuzumab-based therapy; Lauren classification guides chemotherapy selection.
AI systems for gastric cancer pathology have been evaluated at multiple Chinese academic centers including Chinese PLA General Hospital and Peking Union Medical College Hospital. Published studies demonstrate that AI systems for both Lauren classification and HER2 scoring achieve accuracy comparable to senior pathologist consensus while outperforming junior pathologists — particularly significant in provincial hospitals where senior pathology expertise is limited. Processing speed is substantially faster than manual review, enabling same-day results in high-volume centers.
Lung Cancer Subtyping: AI Supporting Treatment Safety
Distinguishing adenocarcinoma from squamous cell carcinoma in NSCLC is not merely diagnostic — it is a treatment safety issue. Certain chemotherapy agents (pemetrexed) and targeted therapies (bevacizumab) are contraindicated in squamous histology due to serious toxicity risk. AI-based NSCLC subtyping directly impacts prescribing safety.
Multiple AI platforms have been validated for NSCLC subtyping at Chinese academic medical centers. Collaborative projects between Chinese hospitals and platforms including PathAI have published multicenter validation studies showing AI subtyping performance that maintains consistency across different scanning hardware — a critical requirement for multi-hospital deployment.
PD-L1 Scoring: Reducing the Variability That Determines Immunotherapy Eligibility
PD-L1 IHC scoring — which determines whether lung, bladder, and other cancer patients qualify for checkpoint immunotherapy — is subject to well-documented inter-observer variability between pathologists. Studies have shown that pathologist disagreement on PD-L1 scoring can change treatment eligibility decisions for a meaningful proportion of patients.
AI PD-L1 quantification systems deployed at Chinese hospitals produce highly consistent scores across repeated measurement and across different readers, reducing the clinical uncertainty that has complicated PD-L1 testing globally. Several NMPA-approved platforms now offer this capability within hospital digital pathology workflows.
Chinese-Developed Platforms with NMPA Approval
Multiple Chinese AI pathology companies have achieved NMPA clearance and active hospital deployment:
- Diano Biotechnology (Beijing): Cervical cytology AI, NMPA-approved 2021
- Panovue (Beijing): Digital pathology whole-slide image platform with integrated cancer analysis AI
- Mindray (Shenzhen): AI hematology analyzer for blood cell morphology, deployed across thousands of hospitals
International platforms from Roche, Leica, and Philips also have significant Chinese hospital installations — creating competitive pressure that accelerates both deployment and quality standards.
Sources: NMPA digital pathology device registry 2025; Chinese PLA General Hospital gastric cancer AI studies; PathAI multicenter NSCLC validation publications; Journal of Pathology (PD-L1 inter-observer variability literature); Diano Biotechnology NMPA approval documentation 2021.