AI PATHOLOGY NSCLC tools may soon help clinicians navigate one of lung cancer care’s most persistent bottlenecks: timely and complete molecular testing. New data show that deep-learning classifiers can accurately identify key driver alterations in non-small cell lung cancer (NSCLC) directly from standard haematoxylin and eosin (H&E)-stained tissue, potentially streamlining treatment selection.
Why Molecular Gaps Still Matter in NSCLC Care
Modern NSCLC management depends on identifying actionable genomic alterations, including EGFR, ALK, BRAF and MET. These biomarkers guide the use of targeted therapies and influence eligibility for neoadjuvant immunotherapy, which is generally considered only in the absence of specific drivers.
Despite clear guidelines, real-world molecular testing remains inconsistent. Limited tissue, long turnaround times and access disparities can result in incomplete profiling, leaving some patients assigned to suboptimal treatment strategies. This has driven interest in AI pathology NSCLC approaches that could complement, rather than replace, molecular diagnostics.
Deep Learning Reads Biomarkers from Routine Slides
In this study, researchers developed AI classifiers using CanvOI 1.1, a digital pathology foundation model trained to analyse H&E-stained tumour sections. Rather than relying on sequencing or immunohistochemistry, the model learned subtle morphological patterns associated with underlying genomic alterations.
Performance was assessed in an independent validation cohort of 968 NSCLC samples. The classifiers demonstrated strong discrimination, with area under the curve (AUC) values of 0.87 for EGFR, 0.96 for ALK, 0.88 for BRAF and 0.83 for MET. Notably, the system also showed high accuracy in identifying tumours lacking these alterations.
Clinician-Friendly Interpretation of AI Outputs
For practising clinicians, the key message is not that AI replaces molecular testing, but that it may function as an intelligent triage tool. Rapid identification of EGFR- or ALK-negative tumours could help prioritise patients for neoadjuvant immunotherapy or expedite confirmatory testing in time-sensitive settings.
Importantly, the high performance for ALK and EGFR, two alterations with major therapeutic implications, suggests clinical utility in flagging likely driver-negative cases when molecular results are delayed or incomplete.
Implications For Real-World Decision-Making
AI pathology NSCLC tools could be particularly valuable in resource-constrained environments or when biopsy material is scarce. By extracting additional value from routine histology, they may reduce delays, support multidisciplinary discussions and improve alignment between tumour biology and treatment choice.
Reference
Rolfo C et al. Validation of histopathology-based deep learning algorithms for detection of actionable non-small cell lung cancer biomarkers. Npj Precision Oncology. 2026; https://doi.org/10.1038/s41698-025-01267-z.
