Research Article
Liver Tumor Segmentation from MR Images Using 3D Fast Marching Algorithm and Single Hidden Layer Feedforward Neural Network
Table 2
Summary of the comparison results.
| Dataset | Evaluation measure | Mean | SD | Min | Max |
| Medic Medical Center | Volumetric overlap error (%) | 26.66 | 7.06 | 15.70 | 39.47 | Percentage volume error (%) | 16.68 | 12.51 | 0.17 | 39.47 | Average surface distance (mm) | 0.44 | 0.47 | 0.21 | 2.12 | RMS surface distance (mm) | 0.97 | 0.97 | 0.53 | 4.40 | Maximal surface distance (mm) | 4.84 | 4.89 | 1.68 | 21.18 |
| TCIA | Volumetric overlap error (%) | 28.57 | 10.89 | 13.71 | 47.89 | Percentage volume error (%) | 14.32 | 15.81 | 0.21 | 47.89 | Average surface distance (mm) | 0.79 | 0.73 | 0.14 | 2.66 | RMS surface distance (mm) | 1.55 | 1.06 | 0.35 | 3.90 | Maximal surface distance (mm) | 8.46 | 6.45 | 1.61 | 19.60 |
| Both | Volumetric overlap error (%) | 27.43 | 8.63 | 13.71 | 47.89 | Percentage volume error (%) | 15.74 | 13.65 | 0.17 | 47.89 | Average surface distance (mm) | 0.58 | 0.60 | 0.14 | 2.66 | RMS surface distance (mm) | 1.20 | 1.03 | 0.35 | 4.40 | Maximal surface distance (mm) | 6.29 | 5.73 | 1.61 | 21.18 |
|
|