Room: Room 207
Purpose: To provide a manufacturer-neutral framework for simulation and reconstruction of low-dose CT patient projection data.
Methods: A current limitation of CT imaging research is the inaccessibility of projection data due to proprietary storage formats used by major manufacturers. To meet the need for accessing non-proprietary projection data, the digital imaging and communications in medicine (DICOM) format was expanded to store CT projection data (CT-PD). This new format, DICOM-CT-PD, has been previously validated, with multiple manufacturers cooperating in providing the necessary data for encoding into this format. To facilitate using the DICOM-CT-PD format to optimize CT scanning protocols or to evaluate novel reconstruction methods, a noise insertion method was implemented to directly simulate reduced dose levels in the DICOM-CT-PD format prior to reconstruction. This method takes into account scanner-specific characteristics such as the bowtie filter, electronic noise level, and automatic exposure control behavior. An interface was created to enable offline CT reconstruction from the DICOM-CT-PD format so that images at multiple dose levels of the same patient exam could be generated. We performed validation on head and body exams, for both phantom and patient data obtained from a Siemens Flash and a GE HD750. Noise measurements taken from the simulated low-dose images were validated against the theoretically predicted noise level, as only routine-dose projection data were available for the patients.
Results: A manufacturer-independent platform was created that enables low dose simulation and reconstruction from a manufacturer-neutral format. Head and body images from two manufacturers were reconstructed with simulated 50% and 25% dose levels, and the difference between the measured and theoretical noise using the inverse square root relation was within an acceptable range of 10%.
Conclusion: A software platform was created to enable low-dose simulation and reconstruction of clinical CT projection data from a manufacturer-neutral format.
Funding Support, Disclosures, and Conflict of Interest: CHM receives research funding support from Siemens Healthineers.