Artificial Intelligence-Advanced Imaging for Solid-Type Lung Adenocarcinoma: Towards Greater Clinical Relevance

Area of Science:

• Oncology
• Radiology
• Artificial Intelligence

Background:

• Solid lung lesions on imaging suggest malignancy but lack a precise definition.
• Subjectivity in defining solid tumors can impact treatment decisions for lung adenocarcinoma.
• Artificial intelligence (AI) may provide objective metrics for tumor characterization.

Purpose of the Study:

• To evaluate AI-based imaging analysis for an objective definition of solid tumors in lung adenocarcinoma.
• To assess AI's ability to predict lymph node metastasis and prognosis.
• To determine the utility of AI in selecting candidates for limited resection.

Main Methods:

• Retrospective analysis of 216 patients with lung adenocarcinoma (≤30 mm).
• AI software calculated consolidation-to-tumour volume (cV/tV) and diameter (cD/tD) ratios.
• Comparison of AI-derived ratios with radiologist assessments and pathological outcomes.

Main Results:

• A cV/tV cutoff of ≥0.72 accurately predicted lymph node metastasis.
• AI analysis improved concordance with radiologists and maintained PathoiD/tD similar to conventional thresholds.
• Tumors with cV/tV <0.72 (≤20 mm) showed no lymph node metastasis and 100% 5-year recurrence-free survival.

Conclusions:

• AI-based volumetric analysis (cV/tV ≥0.72) shows promise for predicting lymph node metastasis in lung adenocarcinoma.
• This AI approach may aid in identifying early-stage tumors suitable for limited resection.
• External validation in multicenter cohorts is crucial for clinical implementation.