Integrating automated three-dimensional body composition with tumor radiomics enhances survival prediction and provides incremental value for postoperative risk stratification in resectable NSCLC.
Both tumor score and body composition score were independently associated with overall survival in the training cohort.
Incorporating body composition radiomics significantly improved discrimination over tumor-only models across all cohorts.
SHAP analysis identified tumor score and body composition score as dominant predictors, stratifying patients into four phenotypes with distinct prognoses.
The study included 1,038 patients across multiple centers with 293 deaths over a median follow-up of 3.31 years.
A fully automated deep learning algorithm was developed for three-dimensional body composition segmentation.
This research suggests that measuring body composition — such as muscle and fat distribution — from CT scans using artificial intelligence can meaningfully improve predictions of how long patients with operable lung cancer will survive after surgery. The study analyzed over 1,000 patients from multiple hospitals, using a deep learning algorithm to automatically measure body composition in three dimensions from imaging scans, then combining those measurements with tumor imaging features and clinical information in a machine learning model. The combined model predicted survival at 1, 2, 3, and 5 years with accuracy above 80% (AUC > 0.80), and performed significantly better than models using tumor imaging alone. This research suggests that the body's physical condition — not just the tumor's characteristics — plays an important role in determining outcomes after lung cancer surgery. By identifying four distinct patient groups based on their tumor and body composition profiles, each with meaningfully different survival rates, the model could help doctors better understand an individual patient's risk. Standard lung cancer staging does not currently account for three-dimensional body composition in an objective, automated way, so this approach represents a potential advancement in personalized prognostic assessment. The practical implication of these findings is that adding automated body composition analysis to routine pre-surgical CT scans — without requiring additional imaging — could help clinicians identify higher-risk patients who might benefit from closer follow-up, earlier supportive care, or more intensive monitoring after surgery. The fully automated nature of the algorithm means this approach could potentially be integrated into existing clinical workflows without requiring manual measurement by radiologists.
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Upload Your LabsHuang Y, Li C, Yang F, Chen X, Chen X, Huang Y, et al.. (2026). Automated three-dimensional radiomic body composition analysis enhances survival prediction in resectable non‑small cell lung cancer.. European radiology experimental. https://doi.org/10.1186/s41747-026-00802-2