VIRTUOS HED file format

The basic format (for versions up to and including 1.2) is as follows:

version         { 1.1 | 1.2 | 1.3 | 1.4 | ... }
modality        { ct | CT | mr | MR | pet | PET | us | US }
created_by      <creator>
creation_info   <comment>
primary_view    { transversal | sagittal | frontal }
data_type       { integer | float | vaxfloat }
num_bytes       { 2 | 4 [ 1 | 8 ] }
byte_order      { aix | AIX | decunix | DECUNIX | linux | LINUX | msdos | MSDOS | sun | SUN | ultrix | ULTRIX | unix | UNIX | vms | VMS  }
patient_name    <name>
slice_dimension <slicedim>
pixel_size      <pixslice>
slice_distance  <slicedist> 
slice_number    <slicenum>
xoffset         <xmin>
dimx            <nx>
yoffset         <ymin>
dimy            <ny>
zoffset         <zmin>
dimz            <nz>

slice_no <slnum_1> position <slpos_1> thickness <slthick_1>
...
slice_no <slnum_n> position <slpos_n> thickness <slthick_n>

z_table         { yes | no }
slice_no  position  thickness  gantry_tilt
<slnum_1> <slpos_1> <slthick_1> <slgtilt_1>
... 
<slnum_n> <slpos_n> <slthick_n> <slgtilt_n>

• version (max. 80 characters)
  describes the file version, interpreted as a floating point number. The version number determines whether some of the following entries are recognized/allowed.
• modality (max. 80 characters)
  describes the imaging modality. The selection list corresponds to CT, MRI, PET and ultrasound, respectively. CT is also used for DOS objects.
• created_by, creation_info (only version>=1.4)
  specify creator and creation comment, respectively. Limited to 200 and 2000 characters, respectively.
• primary_view (80 chars max.)
  specifies the primary view. To the best of my knowledge only transversal is used in our context.
• data_type (only version>=1.2) (80 chars max.)
  specifies the data type (integer or floating point), e.g. for binary files. Floating point numbers are assumed to be in IEEE format, except when vaxfloat is specified, in this case the proprietary format of the (now obsolete) VAX line of computers is assumed. vaxfloat is currently not supported by TRiP98.
• num_bytes (integer value)
  specifies the number of bytes used for the chosen data type. In the original VIRTUOS implementation integers are assumed to be short (2 bytes) and floats are assumed to be single precision (4 bytes). TRiP98, however, supports extensions of one and four byte integers as well as double precision IEEE floats (8 bytes).
• byte_order (80 chars max.)
  describes the endianness of associated binary data. For (presumably) historical reasons this is linked to the operating system, rather than to the endianness itself. So aix, sun and unix denote "big endian", whereas decunix, linux, msdos, ultrix and vms denote "little endian".
• patient_name (79 chars max.)
  specifies the patient name. Note that leading/trailing blanks are significant (except the one blank after the keyword).
  In the GSI pilot project patient names are anonymized and follow the convention Cxynnn where xy are the real patient name's initials and nnn is the running patient number.
• slice_dimension (integer value)
  The meaning of this entry is not quite clear, assume it's the maximum size (voxels) of a data cube in x- and y-direction.
• pixel_size (floating point value)
  Size of a voxel in both x- and y-direction, in mm.
• slice_dist (floating point value)
  Distance between subsequent data slices (voxel size in z-direction), in mm.
• slice_number (integer value)
  The meaning of this entry is not quite clear, assume it's the maximum number of slices (voxels) of a data cube in z-direction. It is also the maximum number of entries in the z_table, if present.
• dimx,dimy,dimz; xoffset,yoffset,zoffset (integer values)
  The meaning of these entries is not quite clear. Assume they describe a subvolume of a binary data cube.
  In the context of TRiP98:

     nx=ny=slicedim
     nz=slicenum
     xmin=ymin=zmin=0
  

• z_table (for version>=1.4), 80 chars max.,
  specifies whether a position list follows or not.
• For position lists, the <slnum_i> denotes an integer slice number, <slpos_i> and <slthick_i> its position and thickness, respectively, in mm, as floating point numbers. The <slgtilt_i> are currently not used within TRiP98. The position list must have <slicenum> entries.

GSI Extensions