Table 2.
Model performance
| Precision | Accuracy | Recall | F1-Score | |
|---|---|---|---|---|
| Resnet (diastole) | 74.08 | 74.61 | 73.93 | 73.93 |
| Resnet (systole) | 77.99 | 77.99 | 77.99 | 77.81 |
| Vit (diastole) | 67.82 | 68.37 | 67.67 | 67.59 |
| Vit (systole) | 75.88 | 76.17 | 76.02 | 75.89 |
| Echo | 72.41 | 73.29 | 57.89 | 85.14 |
| DLFFNet | 82.5 | 82.43 | 91.25 | 82.5 |
| Our model (diastole) | 95.65 | 95.4 | 95.41 | 95.46 |
| Our model (systole) | 93.48 | 93.5 | 93.49 | 93.58 |
| No preprocessing | 92.47 | 92.61 | 93.69 | 92.63 |
| No dynamic adjustment | 96.68 | 96.39 | 96.73 | 96.34 |
| Our model | 98.59 | 96.79 | 97.22 | 97.97 |
Our model (diastole) and (systole) are single-input modes for diastolic and systolic images, respectively. No preprocessing is based on the original model without the image preprocessing process. No dynamic adjustment is based on the original model without the dynamic adjustment function in the SEM