Hi Christian, Thank you for the quick reply. Sorry. I'm a begginer with FSL so I don't know where fslhd and fslstats are. What input should I specify for those scripts? The data has beed preprocessed in SPM5 (realigned and normalised). I smooth them with 6mm kernel in FSL before doing MELODIC. I may do downsampling without further smoothing which might look more or les the same after thresholding. Thanks, Roman Christian Beckmann <[log in to unmask]> wrote: Hi, I'm a bit surprised that it seems to choke on 400 vols on your machine - if you send the output from fslhd and fslstats -V then I can give you a guestimate of how much mem it will require. Is it the case that you've 'normalised' the data first? you can also downsample the effective resolution in space, i.e. have it estimate components from 3x3x3mm voxels rather than 2x2x2mm. You can use flirt for this flirt -in in -ref ref -out out -applyisoxfm 5 wil downsample data set 'in' to 5mm isotropic voxel size. Switching from 16 to 8 bit will not make a difference - internally melodic works on doubles. hope this helps Christian On 14 Aug 2007, at 15:06, Roman Rodionov wrote: > Hi Steve, > > Thank you for the quick reply. > Saying downsampling do you mean changing the number of volumes or > slices? > Would it be enough to switch from 16-bit data to 8-bit? > > Thanks, > > Roman > > > Steve Smith wrote: HI - this will depend on > the 3D matrix size in your data, and how > many non-background voxels there are. > It sounds likely that you are running over the single-process-size > limit for 32-bit machines, in which case you either need to > downsample your data or run on a 64-bit machine. > > Cheers. > > > On 14 Aug 2007, at 12:52, Roman Rodionov wrote: > > > Dear FSLers, > > > > Doing MELODIC for 400 volumes of EPI (400 MB NIFTI 4D file). > > Keep getting the same error irrespective to different settings: > > > > Reading data file filtered_func_data ... done > > Create mask ... done > > Excluding voxels with constant value > > terminate called after throwing an instance of 'std::bad_alloc' > > what(): St9bad_alloc > > > > I don't do any preprocessing (all done bfore combining 3D volumes > > into 4D) > > > > Reduced the number of images downto 50 and it is working. > > > > This has been performed on the mashine with 2GB of RAM showing at > > least 270 > > MB free while MELODICing 50 volumes. Linux. 32 > > > > Could you please let me know rough estimation of the memory > > required for 400 > > volumes processing? I thought it should be enough to have several > > GB of swap > > space. > > Could it be another error (not insufficient memory)? > > > > Thank you, > > > > Roman > > > ---------------------------------------------------------------------- > -- > --- > Stephen M. Smith, Professor of Biomedical Engineering > Associate Director, Oxford University FMRIB Centre > > FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK > +44 (0) 1865 222726 (fax 222717) > [log in to unmask] http://www.fmrib.ox.ac.uk/~steve > ---------------------------------------------------------------------- > -- > --- > > > Yahoo! Answers - Get better answers from someone who knows. Try it > now. ____ Christian F. Beckmann University Research Lecturer Oxford University Centre for Functional MRI of the Brain (FMRIB) John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK. [log in to unmask] http://www.fmrib.ox.ac.uk/~beckmann tel: +44 1865 222551 fax: +44 1865 222717 --------------------------------- Yahoo! Answers - Get better answers from someone who knows. Tryit now.