Hyperconnected Openings Codified in a Max Tree Structure: An Application for Skull-Stripping in Brain MRI T1

Area of Science:

• Medical Imaging
• Computer Vision
• Image Processing

Background:

• Accurate brain segmentation is crucial for neurological studies and clinical applications.
• Existing segmentation methods face challenges with complex anatomical structures and image noise.
• Magnetic resonance imaging (MRI) T1-weighted images are widely used for brain analysis.

Purpose of the Study:

• To develop and evaluate two new procedures for efficient brain segmentation using hyperconnected image processing.
• To leverage novel max-tree based openings for enhanced segmentation accuracy.
• To compare the proposed methods against established brain extraction tools.

Main Methods:

• Implementation of two procedures: maximal hyperconnected function and hyperconnected lower leveling.
• Utilizing new hyperconnected, viscous openings on a max-tree structure.
• Segmentation of 38 T1-weighted MRI datasets from the Internet Brain Segmentation Repository.

Main Results:

• The proposed methods achieved efficient brain segmentation.
• Quantitative evaluation using Jaccard and Dice indices demonstrated competitive performance.
• Comparison with the Brain Extraction Tool and other algorithms showed comparable or improved results.

Conclusions:

• The novel hyperconnected image processing procedures offer an efficient and accurate approach for brain segmentation.
• The max-tree based openings provide a robust framework for segmentation tasks.
• This work contributes a valuable tool for neuroimaging analysis.