Detection of sarcopenic obesity and prediction of long‐term survival in patients with gastric cancer using preoperative computed tomography and machine learning
Problem
Manually measuring body composition from CT scans to identify sarcopenic obesity in gastric cancer patients is slow and has only been studied in small populations; the study assessed its prevalence and prognostic value at scale.
Why it matters
Nutritional status and body composition predict long-term survival after gastric cancer surgery, informing clinical decisions.
What the AI did
Machine learning algorithms automatically detected the third lumbar vertebral level and segmented body composition from preoperative CT images of 840 patients.
Results
Of 840 patients, 63.5% had visceral obesity, 14.2% sarcopenia, and 5.7% sarcopenic obesity; survival results truncated in abstract.
Impact
AI-enabled automated CT analysis made large-scale body composition classification feasible in this study.
Record
| Venue | Journal of Surgical Oncology |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Jae Hyuk Kim, Seung Hee Han, Hyoung‐Il Kim |
| Citations | 34 (OpenAlex, at import time) |
| References | 42 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1002/jso.26668 |
| Open access | hybrid |
| Keywords | sarcopenic obesity, gastric cancer, computed tomography, machine learning, survival |
Importance rating
real but narrower advance
Automated CT body-composition analysis enabled prevalence/prognosis assessment in 840 gastric cancer patients; sarcopenic obesity found to be an independent mortality risk factor. Real but narrower clinical advance.