Tissue-Based Intensity Standardization Technique: Application to the ADNI Multi-Centric Dataset

Nicolas Robitaille¹, Abderazzak Mouiha¹, Simon Duchesne^1,2

Intensity standardization aims at correcting scanner-dependent intensity variations. Most existing techniques aim at matching the input image histogram onto a standard, while we believe the goal should be to match spatially corresponding tissue intensities. In this study, we present STI, a novel technique implementing such approach. We compared STI to an existing histogram-matching method on the ADNI multi-centric dataset by evaluating the mean absolute intensity error (MAE) of the standardized images with respect to the standard, after non-linear registration onto the standard. STI gave better results by showing significantly lower mean MAE for white matter.