Title | ||
---|---|---|
Deep residual contextual and subpixel convolution network for automated neuronal structure segmentation in micro-connectomics |
Abstract | ||
---|---|---|
•We are the first one to utilize deep residual contextual network with sub-pixel strategy to recover the details of neuronal boundaries in EM image stacks, which significantly improves the accuracy of the segmentation.•On the ISBI EM segmentation challenge, the propose method comes out at the top among the leader board and yields Rand score of 0.98788, which is close to the human accuracy values.•The proposed method contributes to the development of connectomics, which provides neurologists with an effective approach to obtain the segmentation and reconstruction of neurons and helps them reduce the burden of manual neurite labeling and validation. |
Year | DOI | Venue |
---|---|---|
2022 | 10.1016/j.cmpb.2022.106759 | Computer Methods and Programs in Biomedicine |
Keywords | DocType | Volume |
Deep learning,Neuronal structure segmentation,Subpixel convolution,Electron microscopy,Micro-Connectomics | Journal | 219 |
ISSN | Citations | PageRank |
0169-2607 | 0 | 0.34 |
References | Authors | |
0 | 6 |