Related Experiment Video
Updated: Oct 19, 2025
09:31
Automated Quantification of Hematopoietic Cell – Stromal Cell Interactions in Histological Images of Undecalcified Bone
Published on: April 8, 2015
11.8K
Automatic segmentation of blood cells from microscopic slides: A comparative analysis
Deponker Sarker Depto1, Shazidur Rahman1, Md Mekayel Hosen1
1Department of Electrical & Computer Engineering, North South University, Bashundhara, Dhaka, 1229, Bangladesh.
Tissue & Cell
|September 23, 2021
Summary
A new diverse dataset, BBBC041Seg, enables robust cell segmentation for microscopic images. This resource aids in developing algorithms for detecting diseases like malaria from imbalanced cell data.
Area of Science:
- Biomedical image analysis
- Computational pathology
- Machine learning for healthcare
Background:
- Deep learning advances have improved automatic cell segmentation on benchmark datasets.
- Existing datasets lack diversity, limiting generalization of current segmentation methods.
- Diverse cell types and imbalanced instances are crucial for real-world applications.
Purpose of the Study:
- Introduce BBBC041Seg, a large, diverse dataset for cell segmentation.
- Facilitate research on few-shot learning for imbalanced cell classes.
- Establish strong baselines for cell segmentation using deep learning and classical methods.
Main Methods:
- Development of the BBBC041Seg dataset, including infected and uninfected cell types.
- Comparative analysis of classical rule-based and state-of-the-art deep learning segmentation methods.
- Evaluation of algorithm performance on diverse and imbalanced cell populations.
Main Results:
- The BBBC041Seg dataset provides a comprehensive resource for cell segmentation research.
- Established baseline performance metrics for various segmentation approaches.
- Highlighted the need for methods robust to class imbalance and diverse cell morphologies.
Conclusions:
- The BBBC041Seg dataset is expected to drive progress in clinically applicable cell segmentation.
- This resource will support the development of automated diagnostic tools for hematological diseases.
- Future research can leverage this dataset for improved malaria detection and other downstream tasks.
