 |
 |
I am currently helping optimize a DTI sequence on a 3T siemens scanner for human work, and also beginning a foray into small animal DTI sequences. I was wondering if anyone could give me some pointers or point me towards some good papers on optimizing sequence parameters (# of echoes, # of averages, # of B0's to collect, # of directions, etc, etc..) Obviously we want the best resolution and best image quality we can get in the shortest amount of time while getting good single to noise and contrast to noise (we're quite ambitious). Just eyeballing the images and saying "looks good" lacks a certain amount of academic rigor, and based on the recent discussion on the list about measuring PHI angles,etc.,etc. simply eyeballing it won't do it. Also the question of what images to even look at-- analyzing all 65+ images by eye is obviously not accurate. Is there a particular way of measuring signal to noise and also contrast to noise, in particular as it applies to tractography? Would looking at the standard deviation of a region outside the brian, and then comparing it to a relatively homologous intracranial region (pick some big hunk of gray matter) be a useful metric? Also do you take a sampling of a certain # of gradients to generate an average SNR across your entire DTI acquisition. Sorry I am relatively naive to all of this, it's just hard to get the sense of what's "good enough". Also can anyone comment on the use of a 30 gradient sequence vs a 64 gradient sequence. Some of my colleagues are trying to integrate DTI in one of their existing protocols and would like it to be as brief as possible, since it's not their main focus. Obviously 30 gradients would take about half as much time as 64 gradients; however I've been using 60+ gradient data now and feel quite compelled to strongly suggest not getting any less than this. Any comments or pointers would be greatly appreciated. DG -- David A Gutman, M.D. Ph.D. Department of Psychiatry & Behavioral Sciences Emory University School of Medicine