Machine Learning–Guided Optical Biopsy Accurately Identifies Malignant Lung Nodules Intraoperatively
Machine Learning–Guided Optical Biopsy Accurately Identifies Malignant Lung Nodules Intraoperatively
Publish Date: 2026-01-15 10:47:00
Source Domain: ascopost.com
-
A cohort study in JAMA Network Open by Azari et al aimed to determine the malignant potential of indeterminate lung nodules during surgery using machine learning-guided intraoperative molecular imaging (IMI) analysis to address surgical challenges in localizing pulmonary nodules.
-
Patients with indeterminate lung nodules treated at the University of Pennsylvania between 2014 and 2021 were included. The study developed a machine learning-based image segmentation algorithm that calculates the tumor-to-background ratio (TBR), integrated with nomograms to create an “optical biopsy” method.
-
The retrospective analysis of 279 patients and prospective study of 61 patients revealed that the machine learning system accurately calculated TBR with reduced variability and processing speed compared to manual methods.
-
The optical biopsy algorithm achieved 96% accuracy in classifying invasive adenocarcinomas in the retrospective validation cohort and displayed a sensitivity of 93.8% and specificity of 100% in a prospective cohort, generating results in under 2 minutes as opposed to the typical 34 minutes for frozen section analysis.
-
The study concluded that AI-analyzed IMI data could reliably determine the malignancy of lung nodules during surgery, potentially improving diagnostic accuracy and real-time surgical decision-making.
