Jun Zhang PhD, Abstract Co-Author: Nothing to Disclose

Nathan C. Hall MD, PhD, Abstract Co-Author: Consultant, Enlyton Ltd Corporation

Michael Vinzenz Knopp MD, PhD, Presenter: Nothing to Disclose

To analyze PET DICOM Tags necessary for SUV calculations and how different implementations can lead to inconsistent SUV readouts

DICOM PET SOP Classes and vendor specific DICOM conformance were reviewed for all major PET and PET/CT systems. Commercial workstations (Siemens LEO, GE AW, Philips EBW and MIMVista) and an in-house developed software were used for SUV comparison. Clinical FDG dosing records were used to compare with dosing info in DICOM headers

Most common pixel units used in commercial PET systems are BQML, COUNTS and GML. GE / Siemens PET systems mainly record pixel in BQML unit while Philips used COUNTS & GML in Allegro/Guardian/Gemini and uses all 3 units in recent Gemini PET. SUV calculations remain identical using BQML among vendors. PET in GML unit leads to SUV directly without decay corrections. Some implementations use private ‘SUV Factor’ to convert COUNTS to SUV. For radiopharmaceutical dose encoding, some populate the DICOM header for injected activity while others store both syringe and residual dose. Substantial inconsistencies by local sites in populating incorrect dosing entry into DICOM were found via a comparison of clinical dosing records versus DICOM header dosing based on 100 PET studies from different environments. Some workstations calculate SUV based on interpolated PET data resulting from either PET/CT fusion or PET alone with variable MPR thickness that cause different SUV readouts. SUV variance between non-interpolated and interpolated PET data mainly depends on interpolation algorithm and data resolution. Most software environments are not able to account for all the variability of the different implementations; some lead to erroneous results

Despite the huge progress in DICOM standardization, variations in PET DICOM implementations and inconsistent SUV readouts between platforms still exists. Rather than focusing on what is wrong, knowing those differences helps with appropriate scaling and adjustment to facilitate consistent data assessment across platforms and further encourage standardization in our community for better consistency of PET/CT examinations across different vendors, devices and software levels

Better understanding variations in PET implementations and such caused SUV inconsistencies between platforms may enable revolution and comparability essential of quantitative clinical applications

Zhang, J, Hall, N, Knopp, M, PET DICOM Tag Implementation for Standardized Uptake Value Calculation: Why SUV Readouts in Workstations Could Be Inconsistent.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8014460.html