A handful of physicists within the Society of Nuclear Medicine and Molecular Imaging (SNMMI) have been engaging industry about what information should be included in SPECT/CT DICOM headers.
Many scanner manufacturers have agreed to make a good faith effort to include all potentially relevant details in the DICOM header if the academic community can coalesce around a common list of information that is needed. Their next meeting with vendors will be concurrent with the EANM annual meeting (October 19-23), and so they would like to present the vendors with a consensus list of DICOM tags ahead of this meeting for their thorough review.
They have taken the liberty of drafting an initial list of details that should be included in DICOM headers for quantitative SPECT, however the community’s input is needed to arrive at the final list that will be presented to the scanner manufacturers.
Your valuable expert input is requested by next Friday (Sept. 20) so that the formed list is provided to industry with ~4 weeks of review time.
Please distribute this post to other colleagues who may have valuable input.
| Brief description of need | Info being requested from scanner manufacturer | Detailed explanation of why the information is needed | Applicable for scanner that provides Bq directly? (Yes, no, explain) |
| Collimator design | Collimator type (LEHR, MEGP, HE, etc) AND hole diameter, septal thickness, and hole length, AND hole shape. Alternatively, if these specifications are public knowledge for a particular vendor’s collimator design, the specific collimator model can be specified. | Reconstructed spatial resolution and scanner sensitivity both depend on the collimator used for image acquisition; in some cases it may be possible for a technologist to utilize the wrong collimator for an imaging procedure. | Yes, but with significantly reduced risk priority number due to the sensitivity factor being controlled for. Residual spatial resolution effects may still impact dosimetry, but <5% for organs. |
| Collimator orbit | Average detector radius throughout acquisition. Ideally would be included for each detector in the DetectorInformationSequence or RotationInformationSequence field. | Reconstructed spatial resolution depends on how close the detectors are to the organs and tumors for which dosimetry is being conducted. Iterative reconstructions are “stopped early” or penalized in a way that prevents complete correction of CDR modeling. This results in increased uncertainty regarding dosimetry measures, particularly for tumors (potentially 10 – 40% difference). For organs, the effect will be <3%. In some cases, the scanners may have not been able to be proximal to patient due to hanging fabric or some other obstruction – in these cases, the detector radius can help as a retrospective troubleshooting tool. | Yes |
| Number of views | Number of projections acquired, and whether it was over 180 deg. or 360 deg | Insufficient projection number can degrade spatial resolution and result in information loss, particularly at the patient periphery. This can cause significant (>20%) dosimetric errors in some cases. This information is also necessary to verify that a particlar protocol was used, for which a validated calibration has been performed. | Yes, unless scanner interlocks acquisition with too few projections, in which interlock failure is the remaining possibility. |
| Acquisition matrix size | Pixel size used for projection acquisition | Insufficient pixel size (<~30% of the intended reconstructed FWHM) can result in information loss and resolution degredation. | Yes, unless it is fixed for all qSPECT acquisitions. |
| Time post-injection | Time elapsed (days, hours, minutes seconds) between end of infusion or start of infusion (PICK ONE OF THESE) and the start of SPECT acquisition. | Allows for verification of decay factor (if used), as well as being a relevant piece of information for dosimetry. | Yes |
| Radiopharmaceutical infusion duration | The time over which the radiopharmaceutical is infused | Long infusion duration may need to be accounted for in dosimetry calculations, particularly with short post-infusion imaging times. | Yes |
| Radiopharmaceutical Information Sequence | All the information about the injection similar as to what is done in a PET DICOM file | Confirmation of radiopharmaceutical being imaged, and the calculated injected activity | Yes |
| Meaning of the units of the projection and reconstructed image | In addition to the units (counts, scaled counts, float counts, etc.), we would want to know whether any corrections being done to the projection data, and how is the reconstructed image rescaled.If any scale factor is being applied, that factor should be specified. | Allows for manual quantitative calculations using a pre-determined system sensitivity factor. | Yes |
| HU to attenuation coefficient mapping for each photopeak | Bi-linear HU to mu mapping for each photopeak. If an “average emission energy” is being used in the case of multiple photopeaks, this should be specified. | Allows for assessment of accurate attenuation correction, as well as whether separate acquisitions/reconstructions may be necessary for individual photopeaks to be corrected separately | Yes |
| Estimated dead time during scan | Average detector dead time during acquisition | Allows for first-order correction of dead time through image scaling. | Yes |
| deadtime correction in the corrections field, as well as the magnitude of correction | Whether dead time correction has already been applied by the scanner manufacturer, and if so, the correction magnitude | Necessary to understand how data was generated for analysis. Can avoid “double-correction” for dead time effects. | Yes |
| Add information for detector swivel angle, detector angle, and the time it spends in each angle | For each detector position and radius, an array of swivel angles and time spent at each angle. | Allows for evaluation of which anatomical regions was emphasized for counting statistics, thereby allowing the end user to better understand lesion detectability vs. position in the body. Also potentially opens up the path for third-party raw data analysis. | Yes |
| Scatter correction method and details of how the method is implemented | Scatter correction method (DEW, TEW, ESSE, MC, other model-based method, etc.), and any tuneable parameters associated with the correction, such as scatter window weights in the DEW/TEW method. | Allows for verification of quantitative nature of imaging data. | Yes |
| Whether or not decay correction during acquisition was included | If decay correction is being applied, is it being applied to a fully reconstructed image, or is it being applied per-projection prior to reconstruction? (I.e. is decay during acquisition handled similar to PET, or not?) | Allows for appropriate post-injection effective imaging timepoint for dosimetry calculations, as well as being potentially relevant for quantitative imaging of isotopes with <10h half-life. | Yes |
| Decay correction – start of scanning or time of injection? | Nature of the decay correction being applied | Knowledge of decay correction method is necessary for SUV calculation, as well as quantitative analysis for dosimetry. | Yes |
| Energy windows acquired, and those used for reconstruction | Energy windows acquired, including scatter windows. | Independent verification of image generation process. | Yes |
| Sensitivity factor (and specify per-head or total) | What is the sensitivity factor used for calculating the quantitative image units (Bq/mL)? | Allows for monitoring system sensitivity over time, as well as quantitative accuracy/constancy. | Only applicable to quantitative scanner. |
| Days since calibration (sensitivity, MHR, uniformity) | Days elapsed since system calibration. This includes MHR, uniformity, sensitivity, and potentially CT HU calibration. | Image quality verification for clinical trials, as well as documentation of QC for adequate clinical performance. | Yes |
| Number of detector heads | Number of detector heads | Manual image quantitation, using a pre-defined sensitivity factor, may be impacted by the number of detector heads in relation to the total number of views acquired. | Yes |
| For images that contain data from multiple photopeaks, attenuation correction performed for each peak or average of peak energies? | – | Assessment of quantitative accuracy of resulting image under varying conditions (i.e. small vs. large patients). | Yes |
| For images that contain data from multiple photopeaks, images added or averaged? | – | ||
| Scanner serial number | Unique identifier to each SPECT or SPECT/CT system. | Need to be able to determine the unique scanner used. Need an identifier. Station Name not always reliable. | Yes |
| When building a protocol, anything the user could change should be in the DICOM header | Any user-modifiable image protocol or reconstruction settings that are not captured elsewhere by the above topics | Verification of image generation process and alignment with scanner performance expectations (both clinical and research). | Yes |
| Time since calibration | Time elapsed from last qSPECT sensitivity calibration | Allows for determination of potential sensitivity drift, as well as potential dose calibrator changes since last calibration date. | Yes |
On behalf of Stephen A Graves, PhD, DABR (University of Iowa) and all members of the SNMMI Working Group for quantitative SPECT DICOM header standardization
Emission Tomography Standardization Initiative (ETSI) 