![IconGems-Print](images/IconGems-Print.png)
375 – Contributed Oral Poster Presentations: Section on Statistics in Imaging
Global Intensity Normalization Induces Correlation in fMRI
Kevin Liu
Marquette University
Daniel B. Rowe
Marquette University/Medical College of Wisconsin
Subject brains vary in both size and physical features. In functional magnetic resonance imaging (fmri), the most intuitive approach to correcting discrepancies between scans are processing with spatial and temporal filtering, whether it be a high or low filter, depending upon your intentions. A less obvious approach is global intensity normalization. As mentioned before, due to the uniqueness of each individual, the mean intensity of an entire data set will vary across individuals and scans. The goal is to retain a constant signal mean across each independent scan. Although this processing induces a correlation of no biological origin, leading to inaccurate interpretations of the data.