Table 1. Summary of CT comparative studies with proposed algorithms and results of EAT and pericardial adipose tissue compartments.
| Authors | AI category | Adipose tissue compartments | Algorithm | Accuracy | Number of patients | Dice score | Level of evidence* |
|---|---|---|---|---|---|---|---|
| Rodrigues et al. [2015] (40) | Machine learning | EAT, mediastinal adipose tissue | J84Graft | 99.00% | 20 | 96.80% | 3 |
| Random Forest | 98.90% | ||||||
| REPTree | 98.90% | ||||||
| J84 | 98.90% | ||||||
| SimpleCart | 98.90% | ||||||
| SMO | 98.30% | ||||||
| RandomTree | 97.50% | ||||||
| RBFNetwork | 96.80% | ||||||
| Spegasos | 96.80% | ||||||
| DecisionStump | 96.80% | ||||||
| HyperPipes | 94.80% | ||||||
| NaiveBayes | 86.00% | ||||||
| Rodrigues et al. [2017] (41) | Machine learning | EAT, mediastinal adipose tissue | Rotation Forest + | – | 20 | – | 3 |
| MLP Regressor | 98.70% | ||||||
| RBF Regressor | 98.60% | ||||||
| MLP Regressor | 98.50% | ||||||
| SMO Regressor | 98.50% | ||||||
| Rotation Forest + | – | ||||||
| Random Forest | 98.20% | ||||||
| Additive Regressor + Random Forest | 98.10% | ||||||
| k-NN/IBk | 98.00% | ||||||
| Random Forest | 97.60% | ||||||
| M5P | 96.90% | ||||||
| Alternating | – | ||||||
| Model Tree | 96.90% | ||||||
| M5 Rules | 96.80% | ||||||
| Linear Regression | 95.30% | ||||||
| Extra Tree | 95.10% | ||||||
| LeastMedSq | 94.90% | ||||||
| Elastic Net | 94.90% | ||||||
| REP Tree | 94.30% | ||||||
| Random Tree | 93.40% | ||||||
| Priyaet al. [2019] (42) | Machine learning | EAT, mediastinal adipose tissue, pericardial adipose tissue | Proposed methodology | EAT 98.5% | 20 | EAT 98.7% | 3 |
| Mediastinal adipose tissue 98.4% | Mediastinal adipose tissue 98.2% | ||||||
| Pericardial adipose tissue 96.4% | Pericardial adipose tissue 98.5% | ||||||
| Commandeuret al. [2018] (43) | Deep learning | EAT, thoracic adipose tissue | ConvNets | EAT 82.3%; thoracic adipose tissue 90.5% | 250 | 95.3%±0.5% obtained for a threshold ts =53.1%±5.1% | 3 |
| Commandeuret al. [2019] (44) | Deep learning | EAT | ConvNets | – | 70 | 87.30% | 3 |
| Bandekaret al. [2006] (45) | Machine learning | Pericardial adipose tissue | Fuzzy affinity-based framework | Pericardial adipose tissue 99.13%±0.38% | 23 | – | 3 |
*, SIGN100: Scottish Intercollegiate Guidelines Network 2019. CT, computed tomography; AI, artificial intelligence; EAT, epicardial adipose tissue.