White Matter and Gray Matter Segmentation in 4D Computed Tomography.

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Radiology

Background:

• Modern Computed Tomography (CT) scanners provide dynamic contrast information for the whole brain, combining functional and anatomical data.
• Accurate soft tissue segmentation, particularly for white matter (WM) and gray matter (GM), is crucial for perfusion analysis and automated detection of cerebral pathologies.

Purpose of the Study:

• To develop and validate an automated method for segmenting WM and GM in contrast-enhanced 4D CT images of the brain.
• To assess the accuracy and feasibility of this segmentation technique for neuroimaging applications.

Main Methods:

• The method involves intracranial segmentation using atlas registration, followed by refinement with a geodesic active contour model.
• Voxel features including intensity, contextual, and temporal information are extracted.
• A support vector machine (SVM) classifier is employed for WM and GM segmentation using the extracted features.

Main Results:

• The automated segmentation achieved high accuracy, with Dice coefficients of 0.81 ± 0.04 for WM and 0.79 ± 0.05 for GM.
• 95% Hausdorff distances were 3.86 ± 1.43 mm for WM and 3.07 ± 1.72 mm for GM.
• Cross-validation on 22 patients demonstrated the robustness and reliability of the segmentation method.

Conclusions:

• Automated segmentation of WM and GM in 4D CT brain images is feasible with good accuracy.
• This method supports advanced quantitative analysis of brain tissue in dynamic contrast-enhanced CT studies.
• The developed technique has potential applications in diagnosing and monitoring neurological conditions.