Table 7.
Selected studies using image-based ML analysis for diagnosis of various cardiomyopathies.
| Publication | Modality | Biomarker | ML technique | Diagnostic | Sample size | Performance |
|---|---|---|---|---|---|---|
| (49) | MRI | Conventional | BN | HCM/DCM/ARV/MYO | 83 | AUC = 0.79 |
| (28) | MRI | Radiomics | RF/LR | HCM | 62 | AUC = 0.95 |
| (52) | MRI | Conventional | RF | MI/HCM/DCM/ARV | 100 | ACC = 0.86 |
| (14) | MRI | Radiomics | SVM | MI/HCM/DCM/ARV | 100 | ACC = 0.92 |
| (53) | MRI | Conventional | RF | MI/HCM/DCM/ARV | 100 | ACC = 0.92 |
| (4) | MRI | Conventional | RF | MI/HCM/DCM/ARV | 100 | ACC = 0.96 |
| (55) | MRI | Deep Learning | VAE | HCM | 737 | ACC = 1.00 |
| (10) | MRI | Conventional | LR | MI/HCM/DCM/ARV | 100 | ACC = 0.94 |
| (57) | MRI | Conventional | LR | HHD/HCM | 224 | ACC = 0.67 |
| (15) | MRI | Radiomics | SVM | HHD/HCM | 224 | ACC = 0.86 |
| (54) | MRI | Deep Learning | CNN | MI/HCM/DCM/ARV | 100 | ACC = 0.78 |
| (9) | MRI | Conventional | SVM/RF | MI/HCM | 45 | ACC = 0.94 |
| (31) | MRI | Conventional | CL | CHD | 60 | ACC = 0.89 |
| (50) | Echo | Conventional | SVM/RF/ANN | HCM/ATHCM | 139 | ACC = 0.91 |
| (33) | Echo | Radiomics | ANN/GA | HCM/DCM | 90 | ACC = 0.95 |
| (18) | Echo | Deep Learning | CNN | HCM/DCM | 927 | AUC = 0.84 |
| (56) | Echo/MRI | Conventional | SVM | DCM | 69 | ACC = 0.94 |
| (26) | Echo | Radiomics | SVM | DCM/ASD | 439 | ACC = 0.98 |
| (37) | Echo | Deep Learning | CNN/GAN | HCM | 772 | ACC = 0.92 |
| (36) | Echo | Deep Learning | CNN | HCM/CA/PH | 14,035 | AUC = 0.93 |