Digital Image Processing Lab

Department of Radiology, University of Michigan


MIAMI Fuse: Mutual Information for Automatic Multimodality Image Fusion

An Application of MI-based Automated Registration:

Functional Magnetic Resonance Images of the Human Brain

Boklye Kim, Thomas Chenevert, John Jonides*, Charles Meyer
Digital Image Processing Laboratory
Department of Radiology, University of Michigan Medical Center
*Department of Psychology, University of Michigan

Functional MR images were acquired using an EPI (Echo Planar Imaging) system available on the GE Signa 1.5 Tesla scanner in the Radiology Department of the University of Michigan Medical Center. Time-series gradient echo planar images of functional response to each activation paradigm (motor or visual stimulation) were acquired. Signal processing was performed using a paired t-statistics test to determine functional significance of voxel intensity changes due to the activation of each paradigm. The resulting functional volume was then mapped onto the anatomical reference image of the same subject acquired in the same session using 3D SPGR. The registered volumes are superimposed to create a brain activation image with functional and anatomic information. The functional images were colorized in red and blue to represent temporal positive and negative correlations, respectively, in the unilateral activation cycle. Volume rendering was performed to create 3D model of functional-anatomic image data.

Motor Stimulation: Finger Tapping Paradigm (Figure 1)

fMRI head images - motor stimulation Figure 1. (13K color, 312x505 gif)

Visual Field Stimulation Paradigm (Figure 2)

fMRI head images - visual stimulation Figure 2. (44K color, 712x489 gif)



The registrations were performed using our "MIAMI Fuse" (Mutual Information for Automatic Multimodality Image Fusion) software as described fully in Meyer, et al..