An active contour model for medical image segmentation with application to brain CT image

Area of Science:

• Medical imaging analysis
• Computational neuroscience
• Biomedical engineering

Background:

• Cerebrospinal fluid (CSF) segmentation in computed tomography (CT) is vital for computer-aided detection (CAD) of acute ischemic stroke.
• Image noise, low contrast, and intensity inhomogeneity present significant challenges in accurate CSF segmentation.
• Existing segmentation methods often struggle with initialization sensitivity and robustness to image artifacts.

Purpose of the Study:

• To develop and evaluate a novel region-based active contour model for robust CSF segmentation in brain CT images.
• To improve the accuracy and reliability of CSF segmentation, particularly in the presence of image noise and intensity variations.
• To assess the model's performance against established segmentation techniques like region-scalable fitting (RSF) and global convex segment (GCS).

Main Methods:

• The proposed method utilizes a region-based active contour model with an energy function incorporating range and space domain kernel functions, and an edge indicator function.
• Minimization of the energy function, optimized via the deepest descent method within a level set framework, automatically identifies target regions with reduced dependence on initial contours.
• Segmentation accuracy was quantified using the overlap rate between segmented results and a reference standard on both synthetic and real brain CT datasets.

Main Results:

• The developed region-based active contour model achieved a 66% average overlap rate in CSF segmentation experiments on 67 brain CT images.
• This significantly outperformed the region-scalable fitting (RSF) model (16% average overlap) and the global convex segment (GCS) model (46% average overlap).
• The method demonstrated effectiveness even with high noise levels and intensity inhomogeneity in the CT images.

Conclusions:

• The region-based active contour model offers accurate CSF segmentation capabilities, even in challenging imaging conditions characterized by noise and inhomogeneity.
• The proposed method shows significant potential for medical image segmentation applications.
• This technique is a promising development for enhancing computer-aided detection (CAD) schemes for acute ischemic stroke in brain CT imaging.