AUCseg

AUCseg performs automatic segmentation of high-grade gliomas in multi-parametric magnetic resonance imaging (MRI) using unsupervised clustering and morphological processing.

Key Features:

• Unsupervised Tumor Segmentation: Segments high-grade gliomas without requiring labeled training data or pre-trained models.
• Gaussian Mixture Model Clustering: Utilizes Gaussian Mixture Model (GMM) clustering for tumor region identification and segmentation.
• Multi-Parametric MRI Integration: Processes T2-FLAIR, post-contrast T1-weighted (T1-CE), and T2-weighted MRI images for tumor component segmentation.
• Tumor Subregion Identification: Separately delineates whole tumor regions, enhancing tumor regions, and necrotic tumor components.

Scientific Applications:

• Glioma Imaging Analysis: Enables segmentation of high-grade gliomas from multi-parametric MRI datasets.
• Neuro-oncology Research: Supports quantitative analysis of tumor structure and subregions in glioma studies.
• Medical Image Processing Research: Provides an unsupervised framework for evaluating tumor segmentation algorithms in MRI data.

Methodology:

AUCseg applies Gaussian Mixture Model clustering to extract whole tumor regions from T2-FLAIR images, segments enhancing tumor regions from post-contrast T1-weighted images, and delineates necrotic regions using morphological processing or clustering on T2-weighted images.