The precision of anatomical normalization in the medial temporal lobe using spatial basis functions

Area of Science:

• Neuroimaging
• Computational anatomy
• Medical image analysis

Background:

• Spatial normalization is crucial for comparing brain structures across individuals.
• Accuracy in localizing homologous regions impacts neuroimaging analyses like voxel-based morphometry.
• Understanding normalization parameter effects is key to reliable results.

Purpose of the Study:

• To assess the accuracy of spatial basis function normalization using anatomical landmarks.
• To determine how normalization parameters influence the precise colocalization of homologous brain regions.
• To evaluate the impact of template selection and regularization on normalization precision.

Main Methods:

• Investigated spatial normalization accuracy using anatomical landmarks.
• Examined effects of basis function count, Bayesian regularization, and template substitution.
• Assessed face validity via landmark colocalization in typically developing children and those with hippocampal pathology.

Main Results:

• Optimal normalization parameters achieved ~1 mm standard deviation for medial temporal lobe landmark colocalization.
• Suboptimal parameters increased landmark colocalization error up to a standard deviation of 3 mm.
• Constrained normalization, prioritizing anatomical likelihood over intensity matching, yielded better precision.

Conclusions:

• Spatial normalization accuracy is highly dependent on parameter selection.
• Careful parameter optimization is essential for precise anatomical localization in neuroimaging studies.
• Findings inform users of voxel-based morphometry for improved analytical reliability.