Brain extraction using the watershed transform from markers.

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Brain extraction is crucial for neuroimaging research, impacting subsequent analysis.
• Existing methods often require extensive preprocessing or prior information.

Purpose of the Study:

• To introduce the marker-based watershed scalper (MBWSS) for robust brain extraction in T1-weighted MR images.
• To demonstrate the efficacy of MBWSS without relying on atlases or shape models.

Main Methods:

• Utilized the Insight Toolkit (ITK) framework for filtering and segmentation.
• Implemented a marker-based watershed transform with aggressive filtering using large kernels.
• Developed a tool named marker-based watershed scalper (MBWSS).

Main Results:

• MBWSS reliably isolates the brain in T1-weighted MR images.
• The method performs accurately across diverse datasets, including paediatric and adolescent cohorts, and intra-surgical scans.
• Successfully adapted MBWSS for macaque brain extraction, showcasing its flexibility.

Conclusions:

• MBWSS offers an efficient and accurate solution for brain extraction in neuroimaging.
• The tool serves as a valuable first step in processing pipelines, achieving high accuracy without prior information.
• Demonstrates the potential of watershed transforms and aggressive filtering in neuroimaging segmentation.