 |
 |
Dear Chao-Gan, I suspect that the scanner is probably using a slightly different gradient waveform (e.g. more agressive ramp sampling) and so it is not possible to have the exact same scanner hardware settings and the pulse sequence settings without modification of the gradient waveforms. However, it is highly unlikely that the effect of the gradient waveforms would be significant for anything that you are doing. Hence I recommend that you keep the pulse sequence parameters the same and simply modify the hardware settings (rise time and max gradient) to values which will allow you to achieve the desired pulse sequence in POSSUM. It shouldn't require too much change, and you can figure out what are the smallest necessary changes by trial and error - just keep checking the consistency with the "consistency check" button. Doing this should allow you to find the acceptable settings quite quickly. Note that the slice profile has no impact on this problem at all. All the best, Mark On 22 Dec 2011, at 18:41, YAN Chao-Gan wrote: > Hi, Mark, > > Thank you very much for your explanation. > > I have contacted our radiologist, and he said the parameter for our Siemens Tim Trio 3T scanner is: gradient rise time: 200 mT/m/ms; peak gradient strength: 40 mT/m per axis. > > So in possum, I specified Maximal gradient strength (T/m) as 0.04; Rise time (s) as 40/200=0.2ms =0.0002s. > > However, I still got a message "Try changing the TRslc value to greater than 0.072250 s". Any idea? Or I need to revise the slice profile? > > Other information: Pixel Bandwidth=2240, Acquisition Matrix=72, 0, 0, 72 (FOV=72*72), then BW(Hz) in Possum should be 2240*72=161280 right? > > Parameters: TE=0.03s, TR=2.5s, TRslice (should be) = 0.0658s, Number of voxels: X=72, Y=72, Z=38; Voxel size=3*3*3; k-space coverage: 100%; Flip angle: 80; Gap: 0.33mm; BW=143360Hz (If correct); Maximal gradient strength: 0.04 T/m; Rize time: 0.0002s (If correct); Slice profile: /usr/local/fsl/data/possum/slcprof > > > Thank you very much for your help! > > Best, > > Chao-Gan > > On Fri, Dec 16, 2011 at 4:34 AM, Mark Jenkinson <[log in to unmask]> wrote: > Hi, > > This problem is likely to be because your scanner has faster > or stronger gradients than the defaults in Possum. If you > go into the "Scanner properties" menu (expand it by clicking > on the triangle) then you can try decreasing the rise time or > increasing the maximum gradient strength. This should > solve that problem. > > As for motion - each package has its own conventions and > so you cannot directly use outputs from SPM in here to get > *exactly* the same motion, but if you know what the parameters > mean (which is a rotation about the x-axis, etc) then you should > be able to order them (and possibly re-centre the translations) > to get something very close, which is normally good enough for > most simulation purposes. > > All the best, > Mark > > > On 15 Dec 2011, at 21:39, YAN Chao-Gan wrote: > >> Hi, Mark, >> >> Thank you very much for your quick response! >> >> I set BW to 143 360Hz, then it Consistency Check said "Try changing TRslc value to greater than 0.0707s. Actually, my TR=2.5s and Slice number is 39. Thus TRslice should be 0.064s. How can I overcome this problem? >> >> I will try on a cluster. >> >> By the way, for the head motion simulation, "4) center of rotations for the mcflirt motion file is in the bottom left corner of the object and for the POSSUM motion file is in the center of the object." How to define "the center of the object"? The center is not the origin of that NIfTI image? And if the realign parameters is aquired by SPM, then the origin is for each time point itself. Is there a way to do the conversions? That means rotate the reference image but not each time point itself. >> >> Thanks! >> >> Chao-Gan >> >> On Thu, Dec 15, 2011 at 4:21 PM, Mark Jenkinson <[log in to unmask]> wrote: >> Hi, >> >> The BW is in Hz, not Hz/pixel. So the value you need to use is 2240 * 64 = 143 360 Hz, >> which is in the range that the warning message suggests. >> >> For the MR parameters, these only depend on field strength, not on scanner. >> They are intrinsic properties of the tissues at a specific field strength. >> So you can just use the values provided if you are at the right field strength. >> >> For whole brain, long time series there is a lot of calculation and it is best >> suited to a cluster. If you had 50 cores then it would around 3 days. >> However, if you do not have a cluster then you can save time by not doing >> whole brain, but just a limited number of slices. Also, the estimated time is >> only approximate - it is based on our server about 4 years ago. So it is worth >> doing a small test and seeing how accurate the estimate is. If you are lucky >> and your machine is substantially faster then it might be quite a lot less time. >> >> All the best, >> Mark >> >> >> On 15 Dec 2011, at 18:59, YAN Chao-Gan wrote: >> >>> Dear all! >>> >>> Currently I am trying to using POSSUM to simulate BOLD signals, however, I experienced some difficulties. Our sequence parameters can not pass POSSUM's consistency check. >>> >>> The problem is the BW (read-out bandwidth, Hz) parameter. For our sequence, BW is 2240 hz/px. When I set BW=2240, TE=0.03s, then POSSUM's consistency check said "Try changing the BW to a value between 82336 and 782120 OR Try changing the TE to greater than 0.91736s." Do you have any idea to set the correct BW? >>> Other parameters: TR=2.5s; TRslice=0.064; Number of voxels X=64, Y=64, Z=39; Voxel size X=3, Y=3, Z=3; FOV=192*192; k-space coverage=100%; flip angle=80; Number of Volumes=2; Gap=0.3mm. >>> >>> Some additional questions: >>> The MR parameters (relaxation times T1, T2*, spin density &rho, and chemical shift value &delta): Could I just use the table provided by possum as MRpar_3T, or I need to specify the values which is specific to our own scanner. Are these parameters varied a lot between scanners? If the latter, how can we calculate them? >>> Run time: POSSUM need 1.4 days for just two volumes!! If we want to simulate 200 volumes with head motion, it need 140 days!!! I don't know why it is so slow. It seems it is very difficult to simulate head motion affected BOLD images for all the subjects. >>> >>> Thanks! >>> >>> Chao-Gan >> >> > >