Research Article
Segmentation of Hyperacute Cerebral Infarcts Based on Sparse Representation of Diffusion Weighted Imaging
Table 2
Segmentation performance comparison between the proposed and four other methods.
| Methods | DC | SEN | SPE | PPV | NPV |
| Proposed | 0.755 ± 0.118 | 0.758 ± 0.149 | 0.999 ± 0.001 | 0.779 ± 0.141 | 0.999 ± 0.001 | FCM | 0.597 ± 0.204 | 0.585 ± 0.221 | 0.999 ± 0.001 | 0.689 ± 0.230 | 0.999 ± 0.001 | FCM_ROI | 0.606 ± 0.201 | 0.505 ± 0.215 | 0.999 ± 0.001 | 0.871 ± 0.156 | 0.999 ± 0.001 | DM | 0.215 ± 0.213 | 0.565 ± 0.346 | 0.983 ± 0.024 | 0.179 ± 0.200 | 0.999 ± 0.001 | DM_ROI | 0.428 ± 0.342 | 0.380 ± 0.321 | 0.999 ± 0.001 | 0.599 ± 0.404 | 0.999 ± 0.001 |
|
|
SEN for sensitivity, SPE for specificity, PPV for positive prediction value, and NPV for negative prediction value.
|