Related Experiment Video
Updated: Apr 16, 2026
10:33
Expedited Radiation Biodosimetry by Automated Dicentric Chromosome Identification ADCI and Dose Estimation
Published on: September 4, 2017
16.8K
Feature selection for the automated detection of metaphase chromosomes: performance comparison using a receiver
Yuchen Qiu1, Jie Song2, Xianglan Lu3
1Center for Bioengineering and School of Electrical and Computer Engineering, University of Oklahoma, 101 David L. Boren Boulevard, Norman, OK 73019, USA.
Analytical Cellular Pathology (Amsterdam)
|March 13, 2015
Summary
This study identifies optimal features for improving metaphase chromosome detection in high throughput microscopy. Feature selection enhances the accuracy of computer-aided detection (CAD) schemes for better clinical analysis.
Area of Science:
- Biomedical imaging
- Microscopy
- Computational pathology
Background:
- Accurate metaphase chromosome detection is crucial for computer-aided detection (CAD) schemes.
- Feature selection significantly impacts CAD scheme accuracy.
- Optimal feature selection is often application-specific.
Purpose of the Study:
- To identify optimal features for enhancing metaphase chromosome detection performance.
- To optimize CAD schemes for high throughput scanning microscopy.
- To compare feature selection methods for image analysis.
Main Methods:
- Acquired 200 bone marrow cell images using high throughput scanning microscopy.
- Applied 9 distinct features individually to classify images.
- Assessed feature performance using receiver operating characteristic (ROC) analysis.
Main Results:
- The number of labeled regions is effective for on-line CAD schemes.
- A combination of four features (labeled regions, average area, average pixel value, standard deviation of region distance/circularity) is recommended for off-line CAD schemes.
- Identified specific features that improve classification accuracy.
Conclusions:
- The study presents an effective feature selection and comparison methodology.
- This approach facilitates the optimization of CAD schemes for high throughput microscopy.
- Provides a foundation for future advancements in automated chromosome analysis.
