Related Experiment Video
Updated: Jan 17, 2026
06:25
Using Computer Vision Libraries to Streamline Nuclei Quantification
Published on: June 6, 2025
625
Automated quantification and feature extraction of nuclei in diffuse large B-cell lymphoma using advanced imaging
Chee Chin Lim1,2, Gei Ki Tang1, Faezahtul Arbaeyah Hussain3,4
1Faculty of Electronic Engineering and Technology, Universiti Malaysia Perlis, Arau, Perlis, Malaysia.
Biomedical Physics & Engineering Express
|September 15, 2025
Summary
Accurate Diffuse Large B-Cell Lymphoma diagnosis is challenging. This study segmented H&E-stained slides, finding nuclear area differences in MYC-positive, MYC-negative, and normal samples, aiding subtyping.
Area of Science:
- Pathology
- Computational Biology
- Medical Imaging
Background:
- Diffuse Large B-Cell Lymphoma (DLBCL) is a common non-Hodgkin lymphoma subtype.
- Accurate DLBCL diagnosis and subtyping are complex, requiring expert analysis.
- Global incidence of DLBCL necessitates improved diagnostic tools.
Purpose of the Study:
- To develop image segmentation and classification methods for DLBCL slides.
- To differentiate MYC-positive, MYC-negative, and normal DLBCL subtypes using morphological features.
- To evaluate the utility of quantitative image analysis in DLBCL subtyping.
Main Methods:
- Utilized a dataset of 108 H&E-stained DLBCL slide images.
- Applied colour deconvolution for nuclei highlighting and watershed algorithm for segmentation.
- Extracted nuclear morphological features (area, perimeter, diameter, circularity) and colour features (LAB, RGB spaces).
Main Results:
- Significant differences (P < 0.05) were observed in nuclear area among DLBCL groups.
- No significant differences were found for nuclear perimeter, diameter, or circularity.
- Significant differences in colour features (Std Dev L, Std Dev RGB) were detected, but not in mean colour values.
Conclusions:
- Nuclear area is a significant morphological feature for differentiating DLBCL subtypes.
- Quantitative colour analysis provides additional discriminatory information.
- Automated image analysis shows promise for improving DLBCL diagnosis and subtyping.
