Accurate identification of tumor oncogene mutations is critical for selecting targeted therapies in non-small cell lung cancer (NSCLC). Current methods, such as tissue biopsies and molecular testing, are invasive, costly, and sometimes unfeasible.
In our latest systematic review and meta-analysis of over 45,000 NSCLC patients from 124 reviewed studies (51 of which were meta-analyzed) predict key driver mutations, including EGFR, ALK, and KRAS. These findings suggest that routine imaging contains valuable phenotypic information that can support personalized testing pathways. Rather than replacing gold-standard genetic tests, AI-based models can act as triage tools to increase pre-test probability—flagging patients most likely to harbor actionable mutations, streamlining the use of molecular assays, and accelerating access to targeted therapies.
This work represents an important step toward integrating AI into precision oncology, transforming standard-of-care imaging into a powerful tool for patient stratification and treatment optimization.
Almudena Fuster-Matanzo, Alfonso Picó-Peris, Fuensanta Bellvís-Bataller, Ana Jimenez-Pastor, Glen J. Weiss, Luis Martí-Bonmatí, Antonio Lázaro Sánchez, David Bazaga, Giuseppe L. Banna, Alfredo Addeo, Carlos Camps, Luis M. Seijo & Ángel Alberich-Bayarri