Research Article

Framework to Segment and Evaluate Multiple Sclerosis Lesion in MRI Slices Using VGG-UNet

Table 3

Overall performance metrics computed for the test images of this study.

CNN scheme2D planeJaccardDiceAccuracyPrecisionSensitivitySpecificity

UNetSagittal82 ± 1.1887 ± 2.6198 ± 0.6491 ± 2.9288 ± 2.7797 ± 1.08
Axial81 ± 2.1488 ± 3.0698 ± 0.7191 ± 3.5288 ± 3.0898 ± 0.51
Coronal85 ± 2.2988 ± 2.0598 ± 0.3291 ± 1.5587 ± 4.1697 ± 0.96

SegNetSagittal82 ± 1.8489 ± 1.1697 ± 0.7291 ± 2.3787 ± 2.8496 ± 1.38
Axial83 ± 3.0689 ± 1.5596 ± 1.2893 ± 2.1688 ± 3.0697 ± 0.52
Coronal85 ± 1.1689 ± 2.7398 ± 0.5191 ± 1.3787 ± 3.3397 ± 0.68

VGG-UNetSagittal83 ± 1.2291 ± 2.4798 ± 0.3891 ± 1.5789 ± 3.0398 ± 0.49
Axial88 ± 1.0692 ± 2.1898 ± 0.1890 ± 2.5593 ± 1.6898 ± 0.52
Coronal89 ± 1.8492 ± 3.1798 ± 0.2293 ± 2.0794 ± 2.4999 ± 0.06

VGG-SegNetSagittal82 ± 2.0590 ± 2.2298 ± 0.2790 ± 2.0789 ± 2.5598 ± 0.28
Axial87 ± 2.5391 ± 2.7198 ± 0.3190 ± 2.3291 ± 2.1698 ± 0.14
Coronal87 ± 3.1591 ± 1.4598 ± 0.1691 ± 2.1792 ± 1.1898 ± 0.23

VGG-UNet (proposed)Sagittal85 ± 3.0592 ± 2.1899 ± 0.0492 ± 2.1795 ± 0.4899 ± 0.03
Axial91 ± 2.2194 ± 1.0599 ± 0.1094 ± 1.2698 ± 0.1299 ± 0.11
Coronal90 ± 2.6293 ± 1.1199 ± 0.0393 ± 1.9497 ± 1.0599 ± 0.08