Related Experiment Video
Updated: Mar 28, 2026
10:59
Reconstruction of 3-Dimensional Histology Volume and its Application to Study Mouse Mammary Glands
Published on: July 26, 2014
15.2K
Breast histopathology image segmentation using spatio-colour-texture based graph partition method.
A D Belsare1, M M Mushrif1, M A Pangarkar2
1Department of Electronics & Telecommunication Engineering, Yeshwantrao Chavan College of Engineering, Nagpur, India.
Journal of Microscopy
|December 29, 2015
Summary
This study introduces an automated method for segmenting cell nuclei in breast histology images, improving the analysis of normal and malignant tissues. The novel approach enhances diagnostic accuracy for pathologists by accurately identifying nuclear arrangements.
Area of Science:
- Digital Pathology
- Computational Biology
- Medical Image Analysis
Background:
- Histopathology image analysis is crucial for diagnosing breast cancer.
- Manual segmentation of nuclear arrangements is time-consuming and subjective.
- Automating this process can significantly aid pathologists.
Purpose of the Study:
- To develop an automated, integrated spatio-colour-texture based graph partitioning method.
- To segment nuclear arrangements in breast histology images (tubules with lumen, solid islands).
- To assist pathologists in analyzing both normal and malignant breast tissue images.
Main Methods:
- Novel similarity-based superpixel generation integrated with texton representation.
- Creation of a spatio-colour-texture map for breast histology images.
- Weighted distance-based similarity measure for graph generation and normalized cuts for segmentation.
Main Results:
- The proposed algorithm successfully segments nuclear arrangements in normal and malignant breast histology images.
- Quantitative evaluation using a ground-truth database (100 images) created by expert pathologists.
- The method demonstrates superior performance compared to existing segmentation techniques.
Conclusions:
- The developed method offers an effective automated solution for breast histology image segmentation.
- It accurately segments nuclear arrangements, aiding in the differentiation of normal and malignant tissues.
- This approach has the potential to improve the efficiency and accuracy of breast cancer diagnosis.
