Related Experiment Video
Updated: May 31, 2026
09:31
Automated Quantification of Hematopoietic Cell – Stromal Cell Interactions in Histological Images of Undecalcified Bone
Published on: April 8, 2015
WBCAtt+: Fine-grained pixel-level morphological annotations for white blood cell images
Satoshi Tsutsui1, Winnie Pang1, Shuting He2
1Rapid-Rich Object Search (ROSE) Lab, School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore.
Medical Image Analysis
|May 28, 2026
Summary
This study introduces WBCAtt+, a new dataset for white blood cell (WBC) image analysis. It offers detailed morphological attributes and cell components, advancing pathology research and enabling explainable AI.
Area of Science:
- Medical Imaging
- Computational Pathology
- Artificial Intelligence in Medicine
Background:
- Microscopic examination of white blood cells (WBCs) is crucial for diagnosing blood disorders.
- Existing WBC image datasets lack detailed morphological annotations used by pathologists.
- This limits the development of advanced diagnostic and analytical tools.
Purpose of the Study:
- To introduce WBCAtt+, a comprehensive dataset for WBC image analysis.
- To provide detailed annotations including 11 morphological attributes and 5 pixel-level cell components.
- To facilitate research in explainable AI for pathology.
Main Methods:
- Developed WBCAtt+, a novel dataset with 113k image-level labels and 10k segmentation maps.
- Created baseline models for attribute recognition and semantic segmentation using the dataset.
- Designed an attribute recognition model incorporating compositional cell structures.
Main Results:
- WBCAtt+ offers the most comprehensive annotations for WBC images to date.
- Baseline models demonstrate the utility of the dataset for attribute recognition and segmentation.
- The compositional attribute recognition model shows improved performance.
Conclusions:
- WBCAtt+ addresses the gap in detailed WBC image annotations.
- The dataset enables advanced applications like explainable AI and counterfactual generation.
- Public availability of the dataset and code promotes further research in computational pathology.
