I am new in DTI preprocessing and our radiologist did not acquire the reverse acquisition which is useful in running TOPUP ( A tool for estimating and correcting susceptibility-induced distortions) and further eddy. A minimum requirement for using ‘topup’ for correcting distortions is that two spin-echo (SE) EPI images with different PE-directions have been acquired. The best is to acquire them with opposing PE-directions (i.e. A→P AND P→A or L→R AND R→L). An SE-EPI image is the same as a b=0 image acquired as a part of a diffusion protocol. Just as for fieldmaps this pair must be acquired at the same occasion as the full diffusion protocol and the subject must not leave the scanner in between and no re-shimming can be done.
Now, can anyone suggest what best can I do for preprocessing? even if I preprocess the data without TOPUP and eddy, how much effective would that be after preprocessing and what would be the possible outcomes/problems caused by due to missing these steps from the preprocessing.
Please help me with this. I am stuck at this point.
thanks in advance
You may want to see my comments on dti sequences and recent comments. An important consideration is whether the acquisition used a classic Stejskal-Tanner (monopolar) sequence or a twice-refocused spin echo (bipolar) sequence was used. The bipolar sequence dramatically reduces spatial distortion, but requires a longer TE resulting in lower SNR. Siemens defaults to the bipolar sequence, while monopolar is the default for the other vendors (though you can change the defaults).
Since the bipolar sequences do not have as much distortion to begin with, Eddy/TOPUP will have less benefit. The advent of Eddy/TOPUP has led many scientists to choose monopolar sequences using two series with reversed phase encoding. This way they can get the best of both worlds: good SNR with post-processing attenuating the spatial distortion.
From my perspective, being unable to undistort your data will reduce your statistical power, meaning you may not find results. However, if you find significant it suggests they are reliable. The perfect can be the enemy of the good. Research scientists might insist on perfect datasets using state-of-the art equipment and head-coils that are unsuitable for many participants (increasing the
Approach score with the NIH) while clinical scientists work with common widely available equipment using rapid sequences that can be used in a clinical workflow to tackle medical questions (increasing the
Significance score with the NIH).