Table 3.
Reliability of CAC-based risk stratification using deep learning-based fully automated software and manual scoring on chest computed tomography.
| Deep learning-based fully automated scoring | |||||
|---|---|---|---|---|---|
| Absent, CAC = 0 | Low, 0 < CAC ≤ 100 | Intermediate, 100 < CAC ≤ 400 | High, 400 < CAC | ||
| Manual scoring (Ref) |
Absent,
CAC = 0 |
62 | 11 | 0 | 0 |
|
Low,
0 < CAC ≤ 100 |
6 | 395 | 2 | 0 | |
|
Intermediate,
100 < CAC ≤ 400 |
0 | 13 | 260 | 1 | |
|
High,
400 < CAC |
0 | 2 | 9 | 152 | |
Kappa, 0.946, 95% Confidence interval = 0.930-0.972, p <0.001, CAC, coronary artery calcium. Dark blue cells represent the patients who were assigned to the same CVD risk category, and lighter blue cells represent those who were assigned to the neighboring risk group.