Requirements for imaging data

In order to be pre-processed by the Data Factory, the imaging data have to meet some requirements, both regarding the images format and the images meta-data.

Images format

  • The images must be full brain scans.
  • The images must be provided either in DICOM or NIFTI format.
  • The images must be high-resolution (max. 1.5 mm) T1-weighted sagittal images.
  • If the dataset contains other types of images (that is not meeting the above description, e.g. fMRI data, T2 images, etc), a list of protocol names used and their compatibility status regarding the above criterion has to be provided.
  • The images must contain at least 40 slices.

Images Meta-data

Imaging meta-data typically are the information contained in the DICOM tags. If the images are provided in NIFTI format or some of the mandatory DICOM tags are missing from the images, those meta-data must be provided in one or several extra file(s) (e.g. JSON, CSV, etc).

Here is a list of DICOM tags we are interested in :

TAG TYPE MANDATORY DESCRIPTION
PatientID LO Yes Patient identifier.
StudyID SH Yes Study identifier. Used to identify a visit. (Unique per dataset or per patient)
SeriesDescription LO Yes Series description. Used to describe the scanning sequence/protocol. Must be stable over time.
SeriesNumber IS Yes Series number. Used to identify a scan run.
InstanceNumber IS Yes Image (slice) identifier.
ImagePosition DS Yes Image (slice) position.
ImageOrientation DS Yes Image (slice) orientation.
SliceLocation DS Yes Slice location.
SamplesPerPixel US Yes Number of samples (planes) per pixel. Usually, 1 for monochrome (gray scale) and palette color images, or 3 for RGB images.
Rows US Yes Number of rows in the image.
Columns US Yes Number of columns in the image.
PixelSpacing DS Yes Distance between the center of each pixel.
BitsAllocated US Yes Number of bits allocated for each pixel sample.
BitsStored US Yes Number of bits stored for each pixel sample.
HighBit US Yes Most significant bit for pixel sample data.
AcquisitionDate DA 1 Acquisition date. We try to use it as the scan date.
SeriesDate DA 1 Series (scan run) date. If AcquisitionDate is missing, we use it as a scan date.
PatientAge AS 2 Patient age at scan date.
PatientBirthDate DA 2 Patient’s birth date.
MagneticFieldStrength DS No Magnetic field strength.
PatientSex LO No Patient gender.
Manufacturer LO No Scanner manufacturer.
ManufacturerModelName LO No Scanner model name.
InstitutionName LO No Institution name.
StudyDescription LO No Study (visit) description.
SliceThickness DS No Slice thickness in mm.
RepetitionTime DS No The period of time in msec between the beginning of a pulse sequence and the beginning of the succeeding (essentially identical) pulse sequence.
EchoTime DS No Time in ms between the middle of the excitation pulse and the peak of the echo produced (kx=0).
SpacingBetweenSlices DS No Spacing between slices in mm (from center to center).
NumberOfPhaseEncodingSteps IS No Total number of lines in k-space in the ‘y’ direction collected during acquisition.
EchoTrainLength IS No Number of lines in k-space acquired per excitation per image.
PercentPhaseFieldOfView DS No Ratio of field of view dimension in phase direction to field of view dimension in frequency direction, expressed as a percent.
PixelBandwidth DS No Reciprocal of the total sampling period, in hertz per pixel.
FlipAngle DS No Steady state angle in degrees to which the magnetic vector is flipped from the magnetic vector of the primary field.
PercentSampling DS No Fraction of acquisition matrix lines acquired, expressed as a percent.
EchoNumber IS No The echo number used in generating this image. In the case of segmented k-space, it is the effective Echo Number.

For more information about the attributes type, have a look at : http://northstar-www.dartmouth.edu/doc/idl/html_6.2/Value_Representations.html.

NOTE: If the ‘MANDATORY’ column contains a number, only one of the tags marked with the same number is needed (e.g. only one of the ‘PatientAge’ and ‘PatientBirthDate’ tags is needed).