Automatic improvement of deep learning-based cell segmentation in time-lapse microscopy by neural architecture search
importance 4/10
ai assisted
Cell Image Analysis Techniques bioimage analysis stage: research
Problem
Deep learning networks for segmenting live cells in time-lapse microscopy are hard and slow to design, requiring expert knowledge.
Why it matters
Accurate cell segmentation underpins biological image analysis, and manual network design is time-consuming and expertise-dependent.
What the AI did
A neural architecture search method automatically finds an optimal deep network for cell segmentation, jointly searching non-repeatable micro architectures to build the macro network.
Results
not stated in source
Impact
The abstract claims a novel NAS-based solution but is truncated before any quantitative results.
Record
| Venue | Bioinformatics |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Yanming Zhu, Erik H. W. Meijering |
| Citations | 20 (OpenAlex, at import time) |
| References | 17 |
| Model or method | Neural architecture search (NAS) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/bioinformatics/btab556 |
| Open access | hybrid |
| Keywords | cell segmentation, neural architecture search, time-lapse microscopy, deep learning |
Importance rating
4
real but narrower advance
Automating cell-segmentation network design saves expert time in biological image analysis; useful tooling advance, no scale stated.