• Yen-Wei Cheng, Hsiao-Wen Chung, Cheng-Yu Chen, and Ming-Chung Chou.
• Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan.
• Eur J Radiol. 2011 Jul 1; 79 (1): 113-7.

ObjectiveTo propose an acquisition scheme suitable for fiber tractography without cerebrospinal fluid contamination while retaining high signal-to-noise ratio.Materials And MethodsDiffusion tensor imaging (DTI) data at b=0 and 1000 s/mm(2) were acquired using conventional spin-echo echo-planar imaging (TR/TE=4000/69, NEX=4, scan time 2:14) plus one additional fluid-attenuated inversion recovery (FLAIR) b=0 set obtained at TI=2200, TR=6000, and NEX=6 (scan time 1:36). The FLAIR images were intensity-adjusted to match the conventional b=0 images according to the white matter signals. Five healthy adults were scanned, with tractography performed from conventional b=1000 and FLAIR b=0 images using the EZ-tracing algorithm. Total traced volumes were statistically compared among conventional DTI, FLAIR DTI, and the "combined DTI" methods at three different slice thicknesses. The fiber consistency index was derived for five repeated scans to assess possible false tracts.ResultsWhole-brain traced volumes from FLAIR DTI and combined DTI were 9.7% (p<0.001) and 17.0% (p<0.0001) greater than from conventional DTI, respectively (paired Student t-test). Fiber consistency was similar for all three acquisition techniques (p>0.2), suggesting insignificant difference in false tracking. Compared with conventional DTI, the increase in total traced volume for FLAIR DTI diminished with decreasing slice thickness, whereas the tractography advantage for combined DTI became increasingly prominent at thinner slices.ConclusionsThe combined DTI technique is capable of improving tractography with higher signal-to-noise ratio at shorter scan time than FLAIR DTI. Its superiority at thin-slice acquisitions makes it particularly suitable for high-resolution clinical applications.Copyright © 2010 Elsevier Ireland Ltd. All rights reserved.

13 keywords...

hide…