Precision or Paradox? AI-Driven Adiposity Imaging in Women With Overweight and Obesity: A Systematic Review and Meta-Analysis.
Asefa Adimasu Taddese, Bjorn T Tam
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis evaluated AI-driven adiposity imaging in women with overweight and obesity, analyzing 38 studies involving 12,815 participants. AI methods showed high segmentation accuracy, with a Dice score of 0.97, and processing times over 99% faster than manual methods. However, extreme between-study heterogeneity limited the reproducibility of adiposity measurements.
Key findings
- AI segmentation accuracy was high, with a Dice score of 0.97 (95% CI: 0.96-0.98).
- AI processing time was significantly faster, averaging 0.05 minutes compared to 46.5 minutes for manual methods (p < 0.001).
- Subcutaneous fat volume averaged 22,536 cm3 with an I2 of 99.9%, indicating extreme heterogeneity.
- 72% of variance in adiposity measurements arose from technical factors rather than anthropometric differences.
Why it matters
Understanding the capabilities and limitations of AI in adiposity imaging is crucial for clinicians, as it may enhance risk stratification in women with overweight and obesity. However, the variability in results highlights the need for standardized protocols to ensure reliable clinical application.
What to keep in mind
The study noted substantial heterogeneity in adiposity measurements due to inconsistent acquisition parameters and segmentation protocols.