New Method for Nucleus Segmentation in Medical Images
Author Information
Author(s): Chang Shulei, Yang Tingting, Yin Bowen, Zhang Jiayi, Ma Liang, Ding Yanhui, Sui Xiaodan
Primary Institution: School of Information Science and Engineering, Shandong Normal University, Jinan, China
Hypothesis
Can a deformable multi-level feature network improve nucleus segmentation accuracy in medical images?
Conclusion
The DMFNet model significantly improves nucleus segmentation accuracy compared to existing methods.
Supporting Evidence
- The DMFNet achieved a mean average precision (mAP) of 37.8% and mean average recall (mAR) of 47.4% on the MoNuSeg 2018 dataset.
- Ablation studies confirmed the effectiveness of the feature extraction and integration modules.
- DMFNet outperformed several state-of-the-art methods in nucleus segmentation tasks.
Takeaway
This study created a new computer program that helps doctors see and understand tiny parts of cells better, which is important for diagnosing diseases like cancer.
Methodology
The study used a deformable multi-level feature network (DMFNet) for nucleus segmentation, integrating feature extraction and mask generation in a one-stage framework.
Limitations
The model may face adaptability issues with new datasets and has high computational resource requirements.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website