Print

Print
Dear Jean-Maxime,
 
I assume that you use multislice epi and that you see alternating dark and light slices. Slice profiles are never perfect, and therefore there is always an overlap between slices if you tell the scanner not to use an interslice gap. Overlapping regions are excited twice within one TR, causing complete signal loss, but only affecting the slices in the second half of the TR in an interleaved sequence. So what you are looking at are (spatially) alternating thick and thin slices.
To solve this problem, you can use sequential scanning with no interslice gap. However,the best thing you can do is use an interslice gap of approximately 20 %, which is usually the minimally required gap to make sure there is no overlap between slices (you can check your scanner specifications). Then you can either use sequential or interleaved scanning, without the crosstalk effect.
 
 
Good Luck,
Serge Rombouts.

Dear member list,

We would like to conduct fMRI experiments using the following parameters: TR=3s, 28 slices, 5mm per slice, with an interleaved acquisition sequence. Using these parameters, we encounter a crosstalk effect, e.g. some variations in the signal intensity between odd and even slices. We are currently considering three potential solutions to reduce the confounding effect:

  1. Increase the TR (to 6 s). Increasing the TR (to 6 seconds with an inter-slice delay)  would significantly reduce crosstalk. However, such an increase in TR comes at the cost of a decrease in statistical power, due to the reduction in a number of volumes in a given time (of course, it would be possible to increase run length, but this would not be too practical given the already rather long experiments length). Furthermore, in some of our protocols, a long TR is just not acceptable.

  2. Acquire volumes in a sequential acquisition instead of in an interleaved acquisition. Such a design would reduce signal intensity due to incomplete recovery of adjacent slices. An alternative solution could be to use a sequential or interleave acquisition with thinner slices (4 mm) with a 1 mm gap between them. This method allows a faster acquisition rate with more volumes per epoch (more than with a 6 sec. TR), but it results in an unpleasant 20% of brain tissue not being scanned.

  3. Correct for the crosstalk effect a posteriori. The correction could be obtained by estimating the ratio of the averaged intensities of odd and even slices, then multiplying the even slices by this ratio. Such a correction, however, decreases the signal/noise ratio.

Have any of you encountered such a problem? If so, what solution(s) have you adopted?

Thank you very much for your responses.

 

Jean-Maxime Leroux

[log in to unmask]

Concierge du Centre de Recherche

Département de Radiologie

Centre Hospitalier de l'Université de Montréal.

1560, rue Sherbrooke Est, Montréal, Québec

H2L 4M1

(514) 812-9915

(514) 890-8000 #28008

(514) 412-7549 (fax)


Téléchargez MSN Explorer gratuitement à l'adresse http://explorer.msn.com