Related Experiment Video
Updated: Jan 7, 2026
09:03
Exploiting Live Imaging to Track Nuclei During Myoblast Differentiation and Fusion
Published on: April 13, 2019
8.6K
AGAFNet: Adaptive Gated Attention Fusion Network for Accurate Nuclei Segmentation and Classification in Histology
Summary
We developed the Adaptive Gated Attention Fusion Network (AGAFNet) for accurate nuclei segmentation and classification in H&E stained histology images, improving cancer diagnosis and research.
Area of Science:
- Digital pathology
- Computational biology
- Medical image analysis
Background:
- Accurate nuclei segmentation and classification in Hematoxylin and Eosin (H&E) stained histology images are crucial for cancer diagnosis, treatment planning, and research.
- Challenges include irregular cell shapes, unclear boundaries, and class imbalance, hindering precise analysis.
Purpose of the Study:
- To propose the Adaptive Gated Attention Fusion Network (AGAFNet), an innovative deep learning model for nuclei segmentation and classification in H&E images.
- To enhance feature representation, selective focus, and information fusion for improved accuracy.
Main Methods:
- AGAFNet features a U-shaped architecture with dedicated decoders for segmentation and classification.
- It integrates three novel attention-based blocks: Channel-wise and Spatial Attention Integration Block (CSAIB), Adaptive Gated Convolutional Block (AGCB), and Fusion Attention Refinement Block (FARB).
Main Results:
- AGAFNet was evaluated on three large-scale datasets: PanNuke, CoNSeP, and Lizard.
- The model demonstrated comparable performance to existing state-of-the-art methods in nuclei segmentation and classification.
Conclusions:
- AGAFNet offers a robust solution for precise nuclei segmentation and classification in H&E stained histology images.
- The proposed attention-based blocks effectively address challenges in digital pathology image analysis.
