Class | Acc. | TPR | FPR | MAE |
|---|
| Â | Â | Â | Â | DSC | MSD | RMS | HD |
|---|
| Â | Â | Â | Â | Â |
mm
|
mm
|
mm
|
|---|
LVC | 0.998 | 1.000 | 0.000 | 0.082 | 0.386 | 0.442 | 1.344 |
| Â | 1.000 | 1.000 | 0.000 | Â | Â | Â | Â |
LVM | 0.051 | 1.000 | 0.001 | 0.268 | 0.510 | 0.547 | 2.127 |
| Â | 1.000 | 1.000 | 0.000 | Â | Â | Â | Â |
RVC | 0.901 | 1.000 | 0.033 | 0.146 | 0.588 | 0.656 | 2.086 |
| Â | 0.997 | 0.997 | 0.000 | Â | Â | Â | Â |
Av. | 0.650 | 1.000 | 0.011 | 0.165 | 0.495 | 0.548 | 1.852 |
| Â | 0.999 | 0.999 | 0.000 | Â | Â | Â | Â |
WH |
0.998
|
1.000
|
0.000
|
0.089
| 0.460 | 0.509 | 1.698 |
| Â |
1.000
|
1.000
|
0.000
| Â | Â | Â | Â |
- CNN segmentations as in Bai et al. [4]. Manual contours were available through Biobank Application 2964. Classes are LV Cavity (LVC), LV Myocardium (LVM), RV Cavity (RVC), an average over the classes (Av.) and a binary segmentation of the whole heart (WH). First row for each class shows the binary classification accuracy for ‘poor’ and ‘good’ segmentations in the Dice Similarity Coefficient (DSC) ranges [0.0 0.7) and [0.7 1.0] respectively. Second row for each class shows the binary classification accuracy for ‘poor’ and ‘good’ segmentations in the Mean Surface Distance (MSD) ranges [>2.0mm] and [0.0mm 2.0mm] respectively. True-positive and false-positive rates are also shown. We report mean absolute errors (MAE) on the predictions of DSC and additional surface-distance metrics: root-mean-squared surface distance (RMS) and Hausdorff distance (HD)
