Field of View Normalization in Multi-Site Brain MRI.

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Multi-site brain Magnetic Resonance Imaging (MRI) analysis is crucial for large-scale neuroimaging studies but faces significant challenges.
• Image acquisition variability across sites, scanners, and protocols complicates preprocessing steps like skull stripping.
• Existing Field of View (FOV) normalization methods often require manual adjustments or ad hoc preprocessing, limiting their generalizability to diverse multi-site datasets.

Purpose of the Study:

• To develop and validate a generic Field of View (FOV) normalization approach for multi-site brain MRI data.
• To assess the impact of generic FOV normalization on the accuracy and consistency of various skull stripping algorithms.
• To provide a publicly available software tool for FOV normalization in neuroimaging research.

Main Methods:

• Developed a generic FOV normalization method applicable to diverse multi-site brain MRI datasets.
• Experimentally validated the method on images from various scanner manufacturers (Philips, GE, Siemens), field strengths (1.0T, 1.5T, 3.0T), and age ranges (0-90 years).
• Evaluated the improvement in skull stripping accuracy and consistency using five established algorithms: FSL's BET, AFNI's 3dSkullStrip, FreeSurfer's HWA, BrainSuite's BSE, and MASS.

Main Results:

• Demonstrated the feasibility of a generic FOV normalization approach across a wide range of multi-site brain MRI data.
• Showed that generic FOV normalization significantly improves the accuracy and consistency of multiple skull stripping algorithms.
• Observed enhanced performance of skull stripping tools when applied to images preprocessed with the proposed FOV normalization method.

Conclusions:

• A generic FOV normalization strategy is effective for harmonizing diverse multi-site brain MRI data.
• This normalization technique enhances the reliability and accuracy of subsequent image analysis, particularly skull stripping.
• The developed FOV normalization software is released to facilitate robust multi-site neuroimaging studies.