Clinical Report: AI Shows Promise for Intraoperative Lung Adenocarcinoma Assessment

Overview

The SuRImage model demonstrated moderate-to-good discrimination in diagnosing and grading lung adenocarcinoma using smartphone photographs of surgical specimens. It outperformed traditional frozen section analysis in several key metrics.

Background

Intraoperative assessment of lung adenocarcinoma is crucial for guiding surgical decisions, particularly in patients with clinical stage IA disease. Traditional frozen section analysis has limitations including time delays and potential inaccuracies.

Data Highlights

Key Findings

• SuRImage achieved 97% sensitivity and 95% accuracy for binary identification of invasive lung adenocarcinoma.
• For the diagnostic task, SuRImage had an average accuracy of 92%, outperforming frozen section analysis which had 85% accuracy.
• In multiclass grading, SuRImage achieved an average accuracy of 89%, compared to 65% for frozen section analysis.
• Grade 1 invasive adenocarcinoma showed lower discrimination in SuRImage, with AUCs of 0.66 to 0.68.
• Frozen section analysis had an overall concordance rate of 87% with final pathology.
• SuRImage maintained robust performance even among patients with ambiguous frozen section findings.

Clinical Implications

The findings suggest that SuRImage could serve as a valuable tool for intraoperative assessment of lung adenocarcinoma, potentially improving diagnostic accuracy. However, further research is needed to determine its effect on surgical decision-making and patient outcomes.

Conclusion

SuRImage shows promise as an innovative approach for intraoperative lung adenocarcinoma assessment, with superior performance metrics compared to traditional methods. Future studies should focus on its clinical impact.

Related Resources & Content

1. Azari et al, JAMA Network Open, 2026 -- Machine Learning–Guided ‘Optical Biopsy’ Accurately Identifies Malignant Lung Nodules Intraoperatively
2. Azari et al, asco ai in oncology, 2026 -- 'Optical Biopsy' Guided by Machine Learning Identifies Malignant Lung Nodules Intraoperatively
3. Valter et al, The ASCO Post, 2025 -- External Validation Confirms Ability of AI Model to Stratify Recurrence Risk in Early-Stage Lung Cancer
4. American College of Chest Physicians, CHEST, 2025 -- CHEST Releases Guideline on Management of Early-Stage Non-Small Cell Lung Cancer
5. PubMed, 2025 -- Impact of Frozen Section Pathology Examination of Surgical Margins in Sublobar Pulmonary Resections for Clinical Stage IA Non-small Cell Lung Cancer
6. The ASCO Post — External Validation Confirms Ability of AI Model to Stratify Recurrence Risk in Early-Stage Lung Cancer
7. CHEST Releases Guideline on Management of Early-Stage Non-Small Cell Lung Cancer - American College of Chest Physicians
8. Impact of Frozen Section Pathology Examination of Surgical Margins in Sublobar Pulmonary Resections for Clinical Stage IA Non-small Cell Lung Cancer - PubMed