COMPARISON OF NON-UNIFORMITY-CORRECTION ALGORITHMS 
USING HIGH-RESOLUTION T1-WEIGHTED MRI SCANS 
FROM TWO DIFFERENT CENTERS

Figures 11 and 12 compare the performance of five non-uniformity-correction algorithms (spm excluded) on two Minneapolis VAMC and two UCLA high-resolution T1-weighted MRI scans (see text). All non-uniformity- correction algorithms except cma operated on stripped volumes; cma volumes were stripped after bias correction. Principal component analysis was applied to the uncorrected volume and the five bias-corrected volumes for each subject, and scatter plots of the first two scaled eigenvectors (E1, E2) for each subject (volume) are illustrated below. Single slices of the bias-corrected volumes and the extracted bias fields are also shown.

    - Biased Brain Phantom Data (Figs. 1-4)
    - Multiple Repeat T1-weighted MRI Scans of a Single Subject (Figs. 5-10)

    - Return to Introduction Page

Figure 12
Figure 11

Top Row. Scatter plots of the first two scaled eigenvectors (EV1 and EV2) from principal component analyses of uncorrected and five corrected high-resolution UMN MRI volumes acquired ten months apart. Bottom Row. Scatter plots of EV1 and EV2 for a similar analysis of two high-resolution UCLA MRI scans acquired three months apart. The similarity of the 2D patterns can be quantified as follows: in each group (UMN1, UMN2, UCLA1, UCLA2), each method is represented as a 2D vector originating at the center of gravity of the cluster of x,y coordinates. Pairwise similarities between the positions of each method in the plot are then summarized by computing the 6-choose-2 dot products of the vectors for each pair of methods. From these data one can compute the average within-site and between-site correlations, 0.879 and 0.708, respectively.

 

© 2000 by the International Neuroimaging Consortium with funding from the 
Human Brain Project and by the Regents of the University of Minnesota. 
Contact: webmaster@pet.med.va.gov		 The University of Minnesota is an equal opportunity educator and employer.