Atlas-based segmentation for globus pallidus internus targeting on low-resolution MRI

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Parkinson's disease diagnosis and treatment often require precise anatomical localization.
• Accurate segmentation of deep brain structures like the globus pallidus is crucial for surgical planning.
• Low-resolution pre-operative MRI poses challenges for detailed anatomical segmentation.

Purpose of the Study:

• To develop an automated method for segmenting the globus pallidus internal (GPi) on low-resolution pre-operative MRI.
• To improve the accuracy of GPi localization for patients with Parkinson's disease.
• To leverage an ultra-high resolution brain atlas for detailed anatomical mapping.

Main Methods:

• Utilized an ultra-high resolution human brain dataset as an electronic reference atlas.
• Employed landmarks-based rigid registration to align datasets.
• Applied affine and non-rigid surface-based registration to propagate GPi labels onto low-resolution MRI scans.

Main Results:

• Successfully segmented the internal part of the globus pallidus (GPi) on low-resolution MRIs.
• Achieved highly accurate anatomical detail through atlas mapping.
• Demonstrated the utility of the method for precise target localization.

Conclusions:

• The developed automated segmentation method provides accurate anatomical detail for GPi localization in Parkinson's disease patients.
• This technique enhances the precision of pre-operative imaging analysis for neurosurgical interventions.
• The atlas-based registration approach effectively overcomes limitations of low-resolution MRI data.