Research Article
Towards Fine Whole-Slide Skeletal Muscle Image Segmentation through Deep Hierarchically Connected Networks
Table 1
The segmentation results compared with state-of-the-art methods.
| Method | -score () | Precision () | Recall () | FT | DT | FT | DT | FT | DT |
| DC [43] | 48 ± 0.093 | 60 ± 0.138 | 41 ± 0.066 | 54 ± 0.164 | 67 ± 0.194 | 73 ± 0.148 | MCG [44] | 63 ± 0.201 | 71 ± 0.105 | 53 ± 0.136 | 64 ± 0.138 | 80 ± 0.303 | 82 ± 0.091 | DNN-SNM [14] | 76 ± 0.033 | 78 ± 0.080 | 83 ± 0.042 | 85 ± 0.089 | 70 ± 0.058 | 73 ± 0.087 | U-Net [30] | 80 ± 0.143 | 81 ± 0.054 | 87 ± 0.155 | 86 ± 0.076 | 74 ± 0.126 | 77 ± 0.055 | Liu et al. [7] | 82 ± 0.172 | 84 ± 0.061 | 81 ± 0.043 | 84 ± 0.071 | 85 ± 0.202 | 85 ± 0.068 | Our approach | 86 ± 0.184 | 89 ± 0.048 | 91 ± 0.174 | 93 ± 0.050 | 82 ± 0.176 | 86 ± 0.058 |
|
|
σ is the standard deviation.
|