3D Body Composition and Artificial Intelligence-A Novel Tool to Assess Sarcopenia and Predict Postoperative Outcomes in Emergency Abdominal Surgery.

Sarcopenic patients had significantly lower skeletal muscle mass and higher volumes of adipose tissue compared to non-sarcopenic patients.

• Sarcopenic patients had lower skeletal muscle mass (p < 0.001)
• Sarcopenic patients had higher volumes of visceral adipose tissue (p < 0.001)
• Sarcopenic patients had higher volumes of subcutaneous adipose tissue (p < 0.02)
• Sarcopenia was defined based on the lowest quartile for skeletal muscle radiodensity

Sarcopenia was associated with older age in emergency laparotomy patients.

• Sarcopenic patients had a mean age of 73 years versus 57 years in non-sarcopenic patients (p < 0.001)
• 408 patients were included in the retrospective analysis
• Patients were drawn from the Australian and New Zealand Emergency Laparotomy Audit-Quality Improvement (ANZELA-QI) registry at a tertiary Australian hospital from 2018 to 2023

Sarcopenia was associated with increased length of hospital stay following emergency abdominal surgery.

• Sarcopenic patients had a mean length of stay of 26 days versus 15 days in non-sarcopenic patients (p = 0.041)

Sarcopenia was associated with increased ICU admission after emergency abdominal surgery.

• Sarcopenia was significantly associated with intensive care unit admission (p < 0.001)
• This association was identified using AI-derived 3D body composition analysis from multiple CT slices of the lumbosacral region

Sarcopenia was not associated with significant postoperative complications or worse discharge status.

• Sarcopenia was not associated with significant post-operative complications as defined by Clavien-Dindo grade ≥ 3 (p = 0.903)
• Sarcopenia was not associated with worse discharge status (p = 0.138)

The study used a novel AI segmentation model applied to multiple CT slices from the lumbosacral region for 3D body composition analysis, as an alternative to traditional single 2D axial CT slice measurement at the L3 level.

• Traditional sarcopenia measurement uses a single 2D axial CT slice at the L3 level, which is described as time-consuming and providing limited data
• A validated AI segmentation model was used to perform 3D body composition analysis
• Multiple CT slices from lumbosacral regions were used rather than a single slice
• 408 patients underwent this 3D body composition analysis retrospectively