.pdb Format

[ pdb header size, in bytes ] - integer

-- PDB HEADER FOLLOWS --

[ 4x4 transformation matrix which is used to transform from voxel space to scanner space (or some other space) ] - 16 doubles, stored in row-major order (same as VTK).

[ number of pathway statistics ] - integer

for each statistic:
  [ currently unused ] - integer
  [ is this statistic stored per point, or just as an aggregate per path? ] - integer (0 = false, 1 = true)
  [ currently unused ] - integer
  [ name of the statistic ] - char[255]
  [ currently unused ] - char[255]
  [ unique ID - an integer that uniquely identifies this statistic, across files ] - integer

[ number of algorithms ] - integer

for each algorithm:
  [ algorithm name ] - char[255]
  [ comments about the algorithm ] - char[255]
  [ unique ID ] - an integer that uniquely identifies this algorithm, across files ] - integer

[ version number ] - integer (DTIQuery 1.1 = 1, this field did not exist prior to DTIQuery 1.1)

-- PDB HEADER ENDS --

[ number of pathways ] - integer

for each pathway:
  [ header size ] - integer

  -- PATHWAY HEADER FOLLOWS --

  [ number of points ] - integer
  [ algorithm ID ] - integer
  [ seed point index ] - integer, currently unused by DTI-Query

  for each statistic:
     [ precomputed statistical value ] - double

  -- PATHWAY HEADER ENDS --

  for each point:       
     [ position of the point - in 0-indexed voxel space! ] - 3 doubles
  for each statistic:
     IF the statistic is also computed per point (see statistics header, second field):
        for each point:
        [ statistical value for this point ] - double 

for each pathway:
  [ byte offset of pathway header ] - long

-- END OF FILE --