Dear Michael,
please find attached to this email the relevant modified functions for
SPM12b (they are indeed not available from the SPM website yet). They
correspond to the ones that Michael W is referring to in his response
below: with these, you can specify the 'slice order' and 'reference
slice' inputs in unit of time (milliseconds) instead of slice indices;
in this setting, what you enter for TA is not relevant and will not be
used, so you can set it to 0.
Best regards,
Guillaume.
On 30/07/14 12:52, Michael Woletz wrote:
> Dear Michael,
>
> since the TA is only used to get the right spacing between the slices in
> cases where there is a waiting period at the end of the acquisition, it
> is not needed any more when the correct slice times are supplied in the
> multi band version.
> Just set it to any value you want (e.g. 0). The correct TA is already
> implicitly stored in your slice times vector.
>
> If you are using spm_slice_timing.m directly, stick to the TA
> calculation in the description and ignore that you are using multiband
> when specifying the timing vector. For historical reasons, this vector
> is used to calculate the TR and this calculation will only work when you
> ignore that you are using a multiband sequence.
>
> Cheers,
> Michael
>
> On 2014-07-30 13:24, Michael Bannert wrote:
>> Dear Michael and SPM gang,
>>
>> (@Guillaume: I just downloaded SPM12b but I’m afraid the slice timing correction hasn’t been included in the current release yet.)
>>
>> Thanks for your code, Michael! I’m about to use that code on my multiband data and I started to wonder about how to calculate TA (acquisition time) in a multiband scenario where several slices are acquired simultaneously. I have the strong intuition that the suggested formula TA = TR - (TR / nslices) does not apply anymore.
>>
>> I’m working with the following values in my dataset:
>> TR = .870 s
>> nslices = 56
>> MB = 4 (multiband factor: number of slices acquired simultaneously)
>>
>> Now what formula should I use to calculate TA for use with SPM’s slice timing correction algorithm? I had the idea in mind that just dividing the number of slices by the number of slices measured simultaneously should do the trick:
>>
>> This is what the formula what would look like then:
>> TA_MB = TR - (TR * MB / nslices) = .8079 s
>>
>> The old formula for comparison:
>> TA = TR - (TR / nslices) = .8545 s
>>
>> Could somebody please verify that I’m on the right track with the adapted formula?
>>
>> Thank you in advance for your thoughts on this!
>>
>> Cheers,
>> Michael
>>
>> On 02 Jun 2014, at 19:37 , Guillaume Flandin <[log in to unmask]> wrote:
>>
>>> Dear all,
>>>
>>> just to let you know that an implementation of multiband slice timing
>>> correction will make its way shortly in SPM12b. It is a contribution
>>> from Michael Woletz and colleagues from the Medical University of
>>> Vienna, Austria. Their work will be described in a poster at HBM next
>>> week (number 3484).
>>> I haven't looked at your code Michael (thanks for sharing!) but I assume
>>> it should be very similar - with perhaps a difference in how the extra
>>> parameters are specified.
>>>
>>> Best regards,
>>> Guillaume.
>>>
>>>
>>> On 02/06/14 18:02, Michael Erb wrote:
>>>> Hi all,
>>>>
>>>> here my solution to the Multiband Slice timing. I have changed the code
>>>> so that for multi band the slice order is an array where each row give
>>>> the slices which are measured at the same timepoint
>>>>
>>>> % Multi-Band: array
>>>> % [ [ 1 4] same timepoint 1
>>>> % [ 3 6] same timepoint 2
>>>> % [ 2 5] ] same timepoint 3
>>>>
>>>> Hope it works also for you.
>>>>
>>>> Regards,
>>>>
>>>> Michael
>>>>
>>>>
>>>> Am 02.06.2014 15:57, schrieb Torben Lund:
>>>>> That is right. Looking at the code it seems that nslices is rather
>>>>> fundamental for the way the code works, and that extra matlab coding
>>>>> will be needed. You could either rewrite the spm_slicetiming.m code or
>>>>> alternatively split each of the current volumes into (in our example)
>>>>> three volumes, and perform the correction on each of the segments
>>>>> seperately, and then after the correction merge the volumes again. Much
>>>>> better would of course be a slice specific designmatrix in which case
>>>>> this would not be a problem.
>>>>>
>>>>> Best
>>>>> Torben
>>>>>
>>>>>
>>>>> Torben Ellegaard Lund
>>>>> Associate Professor, PhD
>>>>> Center of Functionally Integrative Neuroscience (CFIN)
>>>>> Aarhus University
>>>>> Aarhus University Hospital
>>>>> Building 10G, 5th floor, room 31
>>>>> Noerrebrogade 44
>>>>> 8000 Aarhus C
>>>>> Denmark
>>>>> Phone: +45 7846 4380
>>>>> Fax: +45 7846 4400
>>>>> http://www.cfin.au.dk <http://www.cfin.au.dk/>
>>>>> [log in to unmask] <mailto:[log in to unmask]>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Den Uge:23 02/06/2014 kl. 13.37 skrev Rajat Thomas <[log in to unmask]
>>>>> <mailto:[log in to unmask]>>:
>>>>>
>>>>>> Hi Torben,
>>>>>>
>>>>>> Thank you for your suggestion. I think it is a good compromise because
>>>>>> in this case for example, 1 3 5 which should be offset by zero
>>>>>> (because its acquired simultaneously) but will be offset by TR/15
>>>>>> (using default SPM). Indeed for TRs < 1s, this would be negligible.
>>>>>>
>>>>>> Cheers,
>>>>>>
>>>>>> Rajat
>>>>>>
>>>>>>
>>>>>>
>>>>>> Rajat Mani Thomas
>>>>>> Social Brain Lab
>>>>>> Netherlands Institute for Neuroscience
>>>>>> Amsterdam
>>>>>> ------------------------------------------------------------------------
>>>>>> *From:*Torben Lund <[log in to unmask] <mailto:[log in to unmask]>>
>>>>>> *Sent:*02 June 2014 11:48
>>>>>> *To:*Rajat Thomas
>>>>>> *Cc:*spm Parametric Mapping)
>>>>>> *Subject:*Re: [SPM] Multiband Slice timing
>>>>>> Hi Rajat
>>>>>>
>>>>>> This is a wild guess which I have not testet, but have you tried
>>>>>> a sliceorder vector which reflects the situation e.g. with 15 slices
>>>>>> and multislice acceleration of 3 [1 3 5 2 4 1 3 5 2 4 1 3 5 2 4 ]
>>>>>>
>>>>>> Best
>>>>>> Torben
>>>>>>
>>>>>>
>>>>>> Torben Ellegaard Lund
>>>>>> Associate Professor, PhD
>>>>>> Center of Functionally Integrative Neuroscience (CFIN)
>>>>>> Aarhus University
>>>>>> Aarhus University Hospital
>>>>>> Building 10G, 5th floor, room 31
>>>>>> Noerrebrogade 44
>>>>>> 8000 Aarhus C
>>>>>> Denmark
>>>>>> Phone: +45 7846 4380
>>>>>> Fax: +45 7846 4400
>>>>>> http://www.cfin.au.dk <http://www.cfin.au.dk/>
>>>>>> [log in to unmask] <mailto:[log in to unmask]>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Den Uge:23 02/06/2014 kl. 11.08 skrev Rajat Thomas
>>>>>> <[log in to unmask] <mailto:[log in to unmask]>>:
>>>>>>
>>>>>>> Dear SPMers,
>>>>>>>
>>>>>>> Any suggestion on how to perform multiband slice timing in SPM?
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>> Rajat
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Rajat Mani Thomas
>>>>>>> Social Brain Lab
>>>>>>> Netherlands Institute for Neuroscience
>>>>>>> Amsterdam
>>> --
>>> Guillaume Flandin, PhD
>>> Wellcome Trust Centre for Neuroimaging
>>> University College London
>>> 12 Queen Square
>>> London WC1N 3BG
>>
>> --
>> Michael Bannert, Dipl.-Psych.
>> Vision & Cognition Lab, Centre for Integrative Neuroscience, Univ. of Tübingen & BCCN Tübingen
>> Otfried-Müller-Str. 25, 72076 Tübingen, Germany
>> Phone: +49 - (0)7071 / 29 - 89031
>>
>>
>>
>
> --
> Dipl.-Ing. Michael Woletz
> Medizinische Universität Wien
> Zentrum f. Medizinische Physik und Biomedizinische Technik
> Exzellenzzentrum Hochfeld-MR
> Lazarettgasse 14
> 1090 Wien
> Tel.: +43 1 40400 64650
>
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|