Revisiting intensity normalization in brain [¹⁸F]FDG PET: What is reliable in autoimmune encephalitis?

Area of Science:

• Neuroimaging
• Nuclear Medicine
• Neurology

Background:

• Data-driven intensity normalization is an alternative to traditional methods in brain [18F]FDG PET.
• Autoimmune encephalitis (AE) presents challenges for normalization due to variable metabolic patterns and lack of a reliable disease-free reference region.

Purpose of the Study:

• To compare proportional scaling (PS), iterative PS (iPS), and pons-based reference-region (RR) normalization methods in patients with definite AE.
• To evaluate the effectiveness of different normalization techniques in identifying metabolic abnormalities in AE.

Main Methods:

• Three normalization methods (PS, iPS, RR) were applied to brain [18F]FDG PET images from AE patients and healthy controls (HC).
• Voxel-based comparisons (SPM12) were used to assess differences between AE and HC groups.
• Longitudinal analysis was performed on two limbic AE cases.

Main Results:

• PS and iPS detected basal ganglia and mesiotemporal hypermetabolism in AE vs. HC, which RR failed to identify.
• All methods showed cortical hypometabolism, with RR revealing slightly more extensive areas.
• PS and iPS better captured initial abnormalities in longitudinal follow-up compared to RR.

Conclusions:

• PS and iPS normalization methods appear more robust than RR for brain [18F]FDG PET in AE.
• Standardized normalization protocols are essential for consistent analysis of neuroinflammatory disorders.