894 18 Introduction Results III Diffusion tensor imaging

894. 18 Introduction: Results III: Diffusion tensor imaging (DTI) has been applied to a variety of neurological diseases and brain injuries including mild traumatic brain injuries (MTBI). One of the problems that this technology still faces originates from the fact that it is not clear which one of the various quantities that can be derived from the diffusion tensors is most effective in a given application or diagnosis. Among the scalar quantities in use are fractional anisotropy (FA) and the mean (MD), transversal (Lt) and axial (L 1) diffusivity (see figure 1), all of which are certain functions of the diffusion tensor’s eigenvalues. Here we extend DTI into the temporal domain by comparing these measures obtained from subjects who suffered a concussion in a sequence of scans with control groups who did not suffer from injury. In these comparisons we find changes in the structural integrity of the white matter during the recovery period. Figure 3 shows the number of significantly different voxels in axial slices between the concussed subjects and the first control group for the first (red), second (green) and third (blue) scan. Comparison between the two control groups is shown in black. ØFor all four quantities there are two distinct peaks in the concussed-control comparisons, no such peaks exist in control-control; ØAround Z=115 (cyan region) recovery is seen in MD, Lt and L 1 but not in FA; ØAround Z=96 (purple region) recovery is most pronounced in FA and L 1; Ø No recovery is seen for FA around Z=115 and Lt around Z=96. Figure 5 shows regions where significantly different voxels are found in the first scan only (red), in the first two scans (yellow), in the first and third scan (purple) and in all three scans (white). FA MD ØFor all three scans the different voxels are located in the same regions; ØAt Z=115 only MD shows changes in greater regions bilaterally; ØIn most cases areas shrink from red to yellow to white, indicating recovery. There are few locations where the voxels are different only in the first and third scan but not the second; there are no voxels where there is a difference in the third scan but not in the first. FA Scan 1 Scan 2 Scan 3 Control MD Lt L 1 Z=115 Lt Figure 1: Fractional anisotropy, and mean, transversal and axial diffusivity. For graphical purposes Lt and L 1 are normalized the same way as FA. ØThree college football player who suffered a concussion during the season underwent DTI scans within 24 h of the injury and follow-ups one and two weeks later; ØTwo control groups consisted of five and three non-injured players; Figure 3 Figure 4 shows scaled values for the four quantities in voxels that are different for the three concussed subjects and the averages for the controls. ØFA is significantly increased after concussion in all three scans; ØMD and Lt are decreased; ØExcept for the first scan there is no change in L 1. Single Subjects and Average Over Controls FA S S 2 S Scan 2 Scan 3 1 3 C o n Fig. 2 Lt L 1 Figure 5: Regions with significantly different voxels. ØAll scans were pre-processed and co-registered to a standard template (NMI 152) using FSL [1, 2] and smoothed using a 7 x 7 x 7 mm 3 cubic kernel; ØIn a region of interest where FA exceeded a value of 0. 4 (see fig. 2), t-tests were applied between the two control groups, and the first control and the three scans from the concussed subjects to find those voxels that were significantly different (p<0. 05) in either FA, MD, or transversal or axial diffusivity; MD Z=96 Results II: Methods: ØUsing a color-coding scheme allows for identifying regions where recovery occurs. FA L 1 Figure 4 MD Lt L 1 Conclusions: ØRecovery in the white matter is found in sequences of DTI scans taken within 24 h of a concussion and about one and two weeks later; ØDifferent quantities derived from the diffusion tensors have different sensitivity and dependence on location; ØThe most widely used fractional anisotropy (FA) does not seem to be the best diagnostic measure as it shows differences between the control groups in lower brain areas and no changes in time in the upper cortical regions; ØFurther studies of other quantities, including directional information from the tensors’ eigenvectors, are needed to establish DTI diagnostic technique for detecting MTBI and monitoring recovery from MTBI. References: [1] S. M. Smith et al. (2004) Advances in functional and structural MR image analysis and implementation in FSL. Neuroimage 23: 208 -219 [2] S. M. Smith et al. (2006) Tract-based spatial statistics: Voxelwise analysis of multi-subject diffusion data. Neuroimage 31: 1487 -1505 Acknowledgement: Work supported by NINDS grant 48299 (JASK).