Related Experiment Video
Updated: May 25, 2026
09:17
Digital Spatial Profiling for Characterization of the Microenvironment in Adult-Type Diffusely Infiltrating Glioma
Published on: September 13, 2022
A COMPREHENSIVE FRAMEWORK FOR CLASSIFICATION OF NUCLEI IN DIGITAL MICROSCOPY IMAGING: AN APPLICATION TO DIFFUSE
Jun Kong1, Lee Cooper, Fusheng Wang
1Center for Comprehensive Informatics, Emory University, Atlanta, GA 30322.
Proceedings. IEEE International Symposium on Biomedical Imaging
|January 18, 2012
Summary
This study introduces a framework for classifying glioma nuclei in digital microscopy images, achieving 87.43% accuracy using nuclear and cytoplasmic features. The system supports annotation, data management, and analysis for improved glioma diagnosis.
Area of Science:
- Computational pathology
- Digital image analysis
- Oncology
Background:
- Accurate classification of nuclei in diffuse gliomas is crucial for diagnosis and treatment.
- Existing methods may lack comprehensive support for annotation, data management, and analysis.
Purpose of the Study:
- To develop a comprehensive framework for classifying nuclei in digital microscopy images of diffuse gliomas.
- To integrate human annotation, standardized data management, and efficient data query/analysis capabilities.
Main Methods:
- Annotation of 2770 nuclei from 29 whole-slide glioma biopsy images by neuropathologists.
- Machine-based nuclei segmentation and feature extraction (shape, texture, cytoplasmic staining).
- Standardized data model and spatial relational database for storing features and boundaries.
- Training classifiers using features retrieved via spatial queries.
Main Results:
- Achieved an average classification accuracy of 87.43% across 100 independent five-fold cross-validations.
- Demonstrated the effectiveness of nuclear and cytoplasmic features for classifying six common nuclear classes in gliomas.
- Developed a generic framework adaptable to related applications.
Conclusions:
- The proposed framework effectively supports nuclei classification in diffuse gliomas.
- Nuclear and cytoplasmic features are promising for accurate glioma nuclei classification.
- The framework's generic nature allows for broad applicability in digital pathology.
