Automatic brain extraction and brain tissues segmentation on multi-contrast animal MRI

Area of Science:

• Veterinary neuroscience
• Medical imaging analysis
• Comparative neuroanatomy

Background:

• Brain extraction is crucial for MRI quantification in neuroscience.
• Current tools perform poorly on animal brain images due to human-centric design.
• Accurate animal brain segmentation is vital for veterinary research and diagnostics.

Purpose of the Study:

• To develop and validate an automated, atlas-based algorithm for veterinary brain extraction.
• To assess the algorithm's performance across various animal species, MRI contrasts, and cranial conformations.
• To demonstrate the utility of VIBE in subsequent brain tissue segmentation.

Main Methods:

• Developed an atlas-based Veterinary Images Brain Extraction (VIBE) algorithm.
• Incorporated atlas adaptation and image registration pre-processing steps.
• Validated VIBE using Dice and Jaccard metrics on dog and cat MRI data.

Main Results:

• VIBE achieved excellent brain extraction results, confirmed by high Dice and Jaccard metrics.
• The algorithm demonstrated robustness across T1-weighted, T2-weighted, and T2-weighted FLAIR MRI contrasts.
• Successful application was shown across sagittal, dorsal, and transverse planes, multiple species (dogs, cats), and diverse canine skull shapes.

Conclusions:

• The VIBE algorithm provides an accurate and automated solution for animal brain extraction from MRI.
• Its adaptability to various imaging parameters and animal anatomies makes it a valuable tool for veterinary neuroscience.
• VIBE facilitates improved downstream analyses, such as brain tissue segmentation using K-Means clustering.