Automatic segmentation of subcortical brain structures in MR images using information fusion

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Accurate segmentation of internal cerebral structures is crucial for neurological research and clinical applications.
• Existing methods often struggle with the inherent ambiguity and complexity of anatomical data.

Purpose of the Study:

• To develop and validate a novel automated method for segmenting internal cerebral structures using information fusion.
• To integrate diverse data sources, including magnetic resonance images (MRI) and expert knowledge, for improved segmentation accuracy.

Main Methods:

• A three-step fuzzy logic-based information fusion scheme was employed.
• Morphological, topological, and tissue constitution data were modeled and fused to manage imprecision and uncertainty.
• The method was applied to segment the thalamus, putamen, and head of the caudate nucleus in 14 healthy volunteers using MRI data.

Main Results:

• The automated method demonstrated consistent volume estimation compared to expert quantification and published data.
• High spatial similarity was observed between the computed and manually segmented structures.
• The segmentation accuracy for internal cerebral structures was quantitatively validated.

Conclusions:

• The developed information fusion technique provides a robust and accurate automated method for segmenting internal cerebral structures.
• This generic approach is applicable to segmenting various brain structures defined by expert knowledge and morphological images.
• The method shows promise for advancing neuroimaging analysis and understanding brain anatomy.