Spatial normalization of brain images with focal lesions using cost function masking

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Spatial normalization aligns individual brain images to a standard template.
• Accurate spatial normalization is crucial for lesion localization and group comparisons in neuropsychological studies.
• Automated algorithms can introduce significant distortions in brains with focal lesions.

Purpose of the Study:

• To introduce and evaluate cost-function masking for spatial normalization of brain images with focal lesions.
• To compare the efficacy of cost-function masking against standard normalization methods.
• To propose routine use of cost-function masking for lesion-affected brains.

Main Methods:

• Cost-function masking was implemented to exclude lesion areas from image normalization calculations.
• The technique was evaluated using real brains with focal lesions and simulated lesions in normal brains.
• Performance was compared to standard automated normalization and affine-only normalization.

Main Results:

• Cost-function masking significantly reduced inappropriate image distortion caused by focal lesions.
• The method demonstrated superior performance compared to standard automated normalization and affine-only normalization.
• Normalization accuracy was improved when lesion areas were masked during transformation optimization.

Conclusions:

• Cost-function masking is a superior technique for spatial normalization of brains with focal lesions.
• This method mitigates distortions introduced by automated algorithms in the presence of lesions.
• Routine implementation of cost-function masking is recommended for improved neuroimaging analysis in patient populations.