I haven't tested it, but try changing spm_dicom_convert.m so that the determine_datatype function says: function dt = determine_datatype(hdr) % Determine what datatype to use for NIfTI images be = spm_platform('bigend'); if hdr.BitsStored>16 if hdr.PixelRepresentation dt = [spm_type( 'int32') be]; else dt = [spm_type('uint32') be]; end else if hdr.PixelRepresentation dt = [spm_type( 'int16') be]; else dt = [spm_type('uint16') be]; end end Best regards, -John On 18 January 2012 10:39, Michael Marxen <[log in to unmask]> wrote: > Dear Guillaume, dear John, > > The bad news is that there is a problem with release 4290. I just died the > following: > 1) Dicom conversion under spm8 v4010 -> correct import to int16 > 2) run spm_update(4290) > 3) Dicom conversion under the updated version -> incorrect import to uint16 > > The good news is that I also updated my previous 4290 version with > Guillaume's link to 4624. With this version, > The import works correctly again. > Will there be a new release soon? > > Best, Michael > > > ---------------------------------------------------------------------------- > -------- > Michael Marxen, Ph.D., Dipl.-Phys. > MR Research Physicist and > Technical Co-Director Neuroimaging Centre > > Technische Universität Dresden > Faculty of Medicine Carl Gustav Carus > Department of Psychiatry und Psychotherapy > Section of Systems Neuroscience > > Würzburger Str. 35 > 01187 Dresden > Germany > > Phone: +49 (351) 46342212 > Fax: +49 (351) 46342202 > E-Mail: [log in to unmask] > ---------------------------------------------------------------------------- > -------- > > -----Ursprüngliche Nachricht----- > Von: Guillaume Flandin [mailto:[log in to unmask]] > Gesendet: Dienstag, 17. Januar 2012 15:29 > An: Michael Marxen > Betreff: Re: [SPM] dicom import under version 4290 to data type uint16 > > Dear Michael, > > thanks for this. Could you try the latest in-house version of SPM8: > ftp://ftp.fil.ion.ucl.ac.uk/spm/restricted/beta.Duc0z6iLtE/spm8_r4624.zip > and let me know if everything is all fine with it? > > Best regards, > Guillaume. > > > On 17/01/12 13:57, Michael Marxen wrote: >> Dear SPM-Developers, >> >> I noticed that the latest SPM8 release Dicom import seems to convert >> EPI-images (Siemens Trio) from int16 (signed) to uint16 (unsigned). >> >> I converted the same file with the 4010 release and it converted >> (correctly) to int16. SPM8 is running under Windows7 and MatlabR2010a. >> >> I only noticed this because MRICroN misinterpretes the intensity of >> these uint16 images by a factor of 2. I think a bug in MRICroN. >> >> >> >> However, SPM should also not do such a type conversion, e.g. in case >> dicom data is negative ( I am not sure that the standard allows that >> but most likely it does because our original dicom files are signed > int16. >> ) . It seems safest for an import function not to change data types. >> >> >> >> Is this a bug or intended? >> >> >> >> Best, Michael >> >> >> >> >> >> ---------------------------------------------------------------------- >> -------------- >> >> Michael Marxen, Ph.D., Dipl.-Phys. >> >> MR Research Physicist and >> >> Technical Co-Director Neuroimaging Centre >> >> >> >> Technische Universität Dresden >> >> Faculty of Medicine Carl Gustav Carus >> >> Department of Psychiatry und Psychotherapy >> >> Section of Systems Neuroscience >> >> >> >> Würzburger Str. 35 >> >> 01187 Dresden >> >> Germany >> >> >> >> Phone: +49 (351) 46342212 >> >> Fax: +49 (351) 46342202 >> >> E-Mail: [log in to unmask] >> <mailto:[log in to unmask]> >> >> ---------------------------------------------------------------------- >> -------------- >> > > -- > Guillaume Flandin, PhD > Wellcome Trust Centre for Neuroimaging > University College London > 12 Queen Square > London WC1N 3BG > >