Abstract Archives of the RSNA, 2011
SST16-04
Evaluation of ASiR and Prior Image Constrained Compressed Sensing (PICCS) on Ultra-Low Dose CT Studies
Scientific Formal (Paper) Presentations
Presented on December 2, 2011
Presented as part of SST16: Physics (CT Dose and Reconstruction)
Jie Tang PhD, Presenter: Nothing to Disclose
Sarah J Asuncion, Abstract Co-Author: Nothing to Disclose
Kari A. Pulfer, Abstract Co-Author: Nothing to Disclose
Howard A. Rowley MD, Abstract Co-Author: Research Consultant, Eli Lilly and Company
Research Consultant, W.L. Gore & Associates, Inc
Research Consultant, Medpace, Inc
Research Consultant, H. Lundbeck A/S
Research Consultant, Bayer AG
Research Consultant, General Electric Company
Speaker, Bracco Group
Researcher, Guerbet SA
Guang-Hong Chen PhD, Abstract Co-Author: Research funded, General Electric Company
Research funded, Siemens AG
Research funded, Varian Medical Systems, Inc
Research funded, Hologic, Inc
Several methods have been developed to reduce noise in CT images. Noise reduction often implies the possibility of radiation dose reduction. In this study we quantitatively evaluated the radiation dose reduction performance for both ASiR and an in-house reconstruction technique: PICCS.
Each study contains both a normal dose cerebral CT scan (200 mAs) performed in the standard of care and an ultra-low dose scan (50 mAs). Both scans were acquired for the same human subject under an approved IRB protocol and consent from the subject. The normal dose scans were reconstructed on the HD 750 VCT scanner (GE Healthcare, Waukesha, Wisconsin), while the ultra-low dose scans were reconstructed on the scanner first and then processed using both ASiR (100%ASiR) and PICCS. 17 studies were performed. In each study, three ROIs of different materials (white matter, gray matter, air) were selected to measure CT numbers and noise standard deviations. The full widths at half max (FWHMs) of profiles crossing straight sinus in one study were measured to quantify spatial resolution/image sharpness.
For the white matter ROIs, the mean CT numbers are 41 HU (std 2.6 HU) for normal dose FBP images, 37 HU (std 2.3 HU) for ultra-low dose FBP, 36 HU (std 2.6 HU) for ultra-low dose 100% ASiR, and 37 HU (std 2.5 HU) for ultra-low dose PICCS. For the gray matter ROIs, the measured CT numbers are: 12 HU (std 1.6 HU) for normal dose FBP, 10 HU (std 1.7 HU) for ultra-low dose FBP, 8 HU (std 1.6 HU) for ultra-low dose 100% ASiR, and 10 HU (std 1.8 HU) for ultra-low dose PICCS. The mean noise levels of normal dose FBP are 13, 12, and 10 HU for white matter, gray matter, and air, respectively. In comparison, the corresponding mean noise levels for the ultra-low dose scan are 13, 10, and 10 HU for 100% ASiR and 8, 7, and 6 HU for PICCS for white matter, gray matter, and air, respectively. The image sharpness metric was measured as 1.7 mm (normal dose FBP), 2.4 mm (ultra-low dose 100% ASiR), and 1.8 mm (ultra-low dose PICCS).
The noise level of the quarter-dose 100% ASiR images is close to the normal dose FBP images, but with a sharpness degradation. The noise level of the quarter-dose PICCS images is lower than the normal dose FBP images and 100% ASiR images, but with negligible loss in image sharpness.
Using PICCS, radiation dose can be reduced by 75% or more without sacrificing spatial resolution.
Tang, J,
Asuncion, S,
Pulfer, K,
Rowley, H,
Chen, G,
Evaluation of ASiR and Prior Image Constrained Compressed Sensing (PICCS) on Ultra-Low Dose CT Studies. Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL.
http://archive.rsna.org/2011/11016060.html