Journal of
eISSN: 2376-0060
Submit manuscript...
Review Article Volume 13 Issue 2
Artificial intelligence in lung cancer pathology: tool or paradigm shift?
Enrique Díaz Cantón, Adrián Pablo Huñis
Received: April 24, 2026 | Published: May 4, 2026
Citation: Cantón ED, Huñis AP. Artificial intelligence in lung cancer pathology: tool or paradigm shift? J Lung Pulm Respir Res. 2026;13(2):24-29. DOI: 10.15406/jlprr.2026.13.00340
Abstract
Over the past decade, artificial intelligence (AI) in lung cancer pathology has advanced from proof-of-concept image classifiers to clinically consequential systems that support histological subtyping, mutation inference from hematoxylin and eosin (H&E) wholeslide images, quantification of spatial and immune biomarkers, prognostic modelling, and laboratory workflow redesign. A 2025 prospective silent-trial deployment of a finetuned pathology foundation model for EGFR detection in lung adenocarcinoma showed that an AI-assisted workflow can reduce the number of rapid molecular tests required by a clinically meaningful margin while preserving current clinical-standard performance, providing the strongest real-world evidence so far that computational pathology can alter diagnostic sequencing in routine practice. At the same time, external-validation reviews continue to document methodological fragility, restricted datasets, and inconsistent generalization across scanners, stains, institutions, and specimen types. In this Review, we examine whether AI in lung cancer pathology should still be regarded as a sophisticated diagnostic accessory or whether it constitutes an emerging paradigm shift. Drawing on landmark deep-learning studies, recent systematic and narrative reviews, prognostic and predictive signatures, and the rise of pathology foundation models, we argue that AI has exceeded the definition of an ancillary tool, because it is beginning to change what can be inferred from a histological slide—transforming routine tissue into a high-dimensional computational substrate. However, AI has not yet become a mature paradigm: external validation, governance, interoperability, and workforce integration remain insufficient. Pulmonary pathology in 2026 is therefore best understood as a transitional paradigm in active formation, in which AI-augmented expertise—rather than autonomous algorithms— defines the most realistic and responsible near-term trajectory.
Keywords: artificial intelligence, lung cancer, computational pathology, whole-slide imaging, foundation models, EGFR, biomarker discovery, tumor microenvironment, pathologist-in-the-loop, digital pathology
©2026 Cantón, et al. This is an open access article distributed under the terms of the, which permits unrestricted use, distribution, and build upon your work non-commercially.