Table 3.
Evaluation of the end-to-end performance of the body composition analyses
| Center | Quality control | Fatty muscle fraction | Muscle area (cm2) | Visceral fat area (cm2) | Subcutaneous fat area (cm2) |
|---|---|---|---|---|---|
| A | Passed, n = 82 | 0.009 ± 0.008 (3.1% ± 3.5%) | 3.7 ± 4.1 (2.7% ± 4.4%) | 3.6 ± 4.3 (2.7% ± 3.6%) | 5.4 ± 5.3 (2.7% ± 3.0%) |
| B | Passed, n = 243 | 0.016 ± 0.013 (4.8% ± 4.6%) | 5.4 ± 6.4 (3.5% ± 4.0%) | 3.8 ± 6.2 (3.1% ± 5.0%) | 5.8 ± 11.7 (2.8% ± 4.6%) |
| A | Excluded, n = 1 | 0.046 (9.3%) | 16.0 (16.6%) | 2.0 (2.3%) | 14.9 (10.8%) |
| B | Excluded, n = 35 | 0.033 ± 0.036 (6.1% ± 6.6%) | 18.6 ± 21.6 (14.1% ± 15.6%) | 7.2 ± 10.4 (7.0% ± 8.6%) | 18.4 ± 29.5 (7.8% ± 9.5%) |
Absolute and relative differences (in parentheses) between the values obtained with the proposed pipeline and the manually determined values are listed separately for center A and center B and for all 3D CT scans that were included and excluded by restrictive setting of the tissue segmentation quality control. The excluded cases show markedly lower agreement of muscle area, while FMF agreement is still reasonably good (marked in bold)