The realignment of SPM has been designed for aligning fMRI data, which are
typically about 64x64x48 voxels in size. Because these data are pretty
small, I figured that it would be reasonable to store all the derivatives in
memory as double precision floating point and not worry too much about memory
issues. However, if you have a very large image volume, then memory issues
become a problem. For now, I would suggest either reducing the sample
distance or running it on a 64 bit computer with loads of RAM. In the
meantime, I could go through much of the SPM code changing it so that many of
the computations are done in single precision instead. This will only be for
SPM8 (our current developmental version) though, as we plan to make this
compatible only with later versions of MATLAB (probably 7.1 onwards).
Best regards,
-John
On Thursday 31 January 2008 10:08, Carl Johan Ekman wrote:
> I keep getting this error message when running Estimate Reslice!
>
> Running "Realign: Estimate & Reslice"
>
> Error running job: Out of memory. Type HELP MEMORY for your options.
> In file "E:\Matlab\SPM5\spm5\spm_realign.m" (v1030),
> function "realign_series" at line 271.
> In file "E:\Matlab\SPM5\spm5\spm_realign.m" (v1030),
> function "spm_realign" at line 115.
> In file "E:\Matlab\SPM5\spm5\spm_config_realign.m" (v751),
> function "estwrite_fun" at line 420.
>
>
> I have two coronal 3d spgr series (124 images each) which I have imported
> using Dicom Import resulting i two sAW*.img files which I want to realign
> using Estmate Reslice. (Quality:1, Separation:1, Smoothing(FWHM):2,
> Resliced Images: Mean Image Only, all others default.)
> I am using SPM5, Matlab 7.5 on Windows Vista.
> I have increased virtual memory and swap space to the following:
>
> feature memstats
>
> Physical Memory (RAM):
> In Use: 759 MB (2f7af000)
> Free: 1270 MB (4f611000)
> Total: 2029 MB (7edc0000)
> Page File (Swap space):
> In Use: 1163 MB (48bf2000)
> Free: 6171 MB (181b91000)
> Total: 7335 MB (1ca783000)
> Virtual Memory (Address Space):
> In Use: 550 MB (22618000)
> Free: 2521 MB (9d9c8000)
> Total: 3071 MB (bffe0000)
> Largest Contiguous Free Blocks:
> 1. [at 1c840000] 1289 MB (50960000)
> 2. [at 7fff0000] 1023 MB (3ffae000)
> 3. [at 7c41b000] 50 MB ( 32d5000)
> 4. [at 7174b000] 12 MB ( c25000)
> 5. [at 73546000] 10 MB ( a3a000)
> 6. [at 779f4000] 8 MB ( 8ec000)
> 7. [at 1afa0000] 8 MB ( 8d0000)
> 8. [at 16ad0000] 8 MB ( 800000)
> 9. [at 7f7f0000] 7 MB ( 7f0000)
> 10. [at 18f31000] 7 MB ( 79f000)
> ======= ==========
> 2427 MB (97b8d000)
>
> Does anyone have a solution? Do I need more RAM? Is it possible to
> use "pack" or "clear" in the spm_realign function?
> I would be very grateful for help. I am stuck.
>
> Carl Johan Ekman
> Karolinska Institutet
> Stockholm, Sweden
|