Modified distance regularized level set evolution for brain ventricles segmentation.

Area of Science:

• Medical imaging analysis
• Neuroscience
• Computational anatomy

Background:

• Neurodegenerative disorders often involve brain atrophy, characterized by neuronal loss and altered ventricular morphology.
• Accurate segmentation of brain ventricles is crucial for diagnosing atrophy.
• Existing segmentation methods may struggle with variations in image quality and data types.

Purpose of the Study:

• To present a modified distance regularized level set evolution segmentation method for improved ventricle segmentation.
• To incorporate regional intensity information for enhanced segmentation accuracy.
• To validate the method across multimodal imaging datasets (MRI and CT) for both normal and atrophy subjects.

Main Methods:

• A modified distance regularized level set evolution algorithm was developed.
• Regional intensity information was integrated into the segmentation model.
• The method was applied to segment ventricles in magnetic resonance imaging (MRI) and computed tomography (CT) images.
• Parameter optimization was performed for different datasets.

Main Results:

• The proposed method achieved high performance metrics: sensitivity (65%-90%), specificity (98%-99%), and accuracy (95%-98%).
• Quantitative evaluation using Peak Signal to Noise Ratio (95%) and Structural Similarity Index (0.95) confirmed segmentation accuracy.
• The method demonstrated robustness against noisy images and adaptability to multimodal data.

Conclusions:

• The developed segmentation method is efficient, robust, and adaptive for analyzing brain ventricle morphology.
• This technique aids in the diagnosis of brain atrophy by providing accurate ventricle segmentation.
• The multimodal and adaptive nature of the method supports its broad applicability in neuroimaging research and clinical settings.