Segmentation techniques for the classification of brain tissue using magnetic resonance imaging

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Medical Image Analysis

Background:

• Accurate segmentation of brain tissue components is crucial for neurological research and clinical diagnosis.
• Existing methods may face challenges in reproducibility and accuracy across different imaging platforms.

Purpose of the Study:

• To develop and validate a novel technique for classifying brain tissue into gray matter, white matter, and cerebrospinal fluid.
• To assess the reliability and reproducibility of this classification method using magnetic resonance imaging (MRI) data.

Main Methods:

• Utilized simultaneously registered proton density and T2-weighted MRI images.
• Employed a linear discriminant function trained on identified tissue samples (training classes) for pixel classification.
• Investigated the impact of training class location and number, settling on six pairs for optimal results.

Main Results:

• Achieved good interrater and test-retest reliability for the brain tissue classification technique.
• Demonstrated good intrascanner reproducibility but poor interscanner reproducibility.
• Validated the method through correlation analyses with region of interest segmentation, age, phantom studies, and visual inspection.

Conclusions:

• The developed MRI-based technique offers a valid and reliable method for segmenting gray matter, white matter, and cerebrospinal fluid.
• While reproducible within the same scanner, interscanner variability necessitates further investigation for broader clinical application.
• The method shows promise for various neuroimaging applications, with discussed strengths and limitations.