Groupwise registration of infant brain diffusion tensor images using intermediate subgroup templates.

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Developmental Neuroscience

Background:

• Infant brain development involves rapid changes, complicating image registration.
• Diffusion tensor images (DTI) offer consistent properties for infant brain registration compared to T1/T2-weighted images.
• Existing groupwise registration methods often use scalar images, missing microstructural orientation information crucial for accuracy.

Purpose of the Study:

• To develop and validate an advanced groupwise registration method for infant tensor images.
• To leverage microstructural orientation information from tensor images for improved registration accuracy.
• To reduce deformation and bias in infant brain image alignment using intermediate subgroup templates.

Main Methods:

• Proposed an intermediate subgroup tensor template-based groupwise (IST-G tensor) registration approach.
• Clustered tensor images into subgroups using Louvain clustering based on image similarity.
• Generated subgroup tensor templates and aligned them to a sample-specific common space using DTI-toolkit.

Main Results:

• The IST-G tensor approach significantly enhanced global and local registration accuracy.
• Clustering by image similarity outperformed no clustering and matched chronological age clustering.
• The method effectively utilized tensor map consistency and reduced deformation via intermediate templates.

Conclusions:

• The IST-G tensor registration framework provides more accurate alignment of longitudinal infant brain tensor images.
• Leveraging tensor image properties and subgroup templates improves registration outcomes in early infancy.
• This method facilitates more precise analysis of infant brain development using DTI.