From: Deep learning to diagnose cardiac amyloidosis from cardiovascular magnetic resonance
All n = 206 | Training subgroup n = 134 (65%) | Validation subgroup n = 30 (15%) | Test subgroup n = 42 (20%) | p-value | |
---|---|---|---|---|---|
Age (years) | 76 (69–81) | 77 (69–81) | 71 (65–79) | 73 (63–79) | 0.097 |
Male sex (n, %) | 137 (67) | 88 (66) | 20 (67) | 29 (69) | 0.623 |
BSA (m2) | 1.84 (1.74–1.96) | 1.83 (1.73–1.95) | 1.88 (1.75–2.02) | 1.85 (1.76–1.95) | 0.206 |
Amyloidosis (n, %) | 107 (52) | 71 (53) | 15 (50) | 21 (50) | 0.645 |
AL/ATTR/no amyloidosis (n, %) | 50/57/99 (24/28/48) | 33/38/63 (25/28/47) | 6/9/15 (20/30/50) | 11/10/21 (26/24/50) | 0.610 |
CMR findings | |||||
LVEDVI (mL/m2) | 79 (67–96) | 79 (64–95) | 71 (66–92) | 84 (70–106) | 0.105 |
LVESVI (mL/m2) | 33 (23–50) | 31 (23–50) | 27 (20–42) | 40 (26–56) | 0.112 |
LVSVI (mL/m2) | 44 (37–51) | 42 (37–49) | 44 (39–52) | 47 (38–54) | 0.131 |
LVEF (%) | 55 (46–67) | 56 (47–67) | 62 (51–70) | 53 (45–64) | 0.277 |
CO (L/min) | 5.3 (4.3–6.6) | 5.1 (4.3–6.3) | 5.5 (5.1–7.3) | 5.5 (4.4–7.0) | 0.565 |
LVMI (g/m2) | 84 (71–111) | 88 (71–114) | 82 (70–113) | 82 (73–107) | 0.730 |
LA area index (cm2/m2) | 16 (13–18) | 15 (13–18) | 17 (13–18) | 16 (14–18) | 0.714 |
RA area index (cm2/m2) | 13 (11–15) | 13 (11–15) | 13 (10–17) | 13 (12–15) | 0.742 |
RVEDVI (mL/m2) | 70 (57–81) | 66 (55–80) | 71 (55–89) | 73 (62–81) | 0.212 |
RVESVI (mL/m2) | 28 (20–37) | 28 (20–37) | 27 (19–34) | 30 (22–37) | 0.180 |
RVSVI (mL/m2) | 39 (33–47) | 38 (33–44) | 40 (34–52) | 42 (34–51) | 0.624 |
RVEF (%) | 59 (47–65) | 59 (49–64) | 60 (49–67) | 60 (47–65) | 0.468 |
Early darkening (n, %) | 40 (19) | 26 (19) | 8 (27) | 6 (14) | 0.241 |
LGE presence (n, %) | 174 (84) | 115 (86) | 25 (83) | 34 (81) | 0.954 |
LGE subendocardial-to-transmural pattern (n, %) | 89 (43) | 61 (46) | 11 (37) | 18 (43) | 0.434 |
Pericardial effusion (n, %) | 48 (23) | 32 (22) | 4 (13) | 12 (29) | 0.465 |
Pleural effusion (n, %) | 58 (28) | 39 (27) | 7 (23) | 12 (29) | 0.606 |