Click here to

Session: Best Poster Competition [Return to Session]

Evaluation of Siemens SOMATOM CT-Simulator Reconstruction Space And Correlation Of Physical and Clinical Image Quality

P Patel, D Olek, E Butler, R Pino, Methodist Hospital, Houston, TX

Presentations

(Saturday, 3/26/2022)   [Central Time (GMT-5)]

Purpose: Evaluate the impact of CT-Simulator reconstruction settings on image quality focusing on reconstruction kernel, slice thickness, and strength of Sinogram Affirmed Iterative Reconstruction (SAFIRE).

Methods: A Catphan 504 phantom was scanned on a Siemens SOMATOM Definition Edge CT. The raw image data was reconstructed with each unique combination of reconstruction kernel (Br38, Bv66, Hr38, Hv49, Qr40, Qv43), slice thickness (0.5,1,1.5,2 mm), and SAFIRE Strength (0,1,3,5). Image sets were evaluated with DoseLab Pro v6.80 for spatial resolution, contrast-noise ratio, contrast and noise. Six reconstruction combinations were chosen to evaluate impact on two clinical CT image data sets, head/neck and pelvis. Regions of interest were assessed within specific organs in order to quantify the amount of noise reduction and HU value as compared to the current clinical protocol.

Results: Qr40 provided the lowest noise levels with negligible HU impact based on Catphan acceptable ranges, displaying a noise reduction of 32.1-43.6% over Br38 in the phantom study. This was at SAFIRE Strength 5 with varying slice thickness. Across all reconstruction parameters: spatial resolution ranged from 0.34-0.62mm, contrast-noise from 7.5-66.6, contrast from 26.6-30.2%, and noise from 0.404-3.547%. In the subsequent anatomy ROI study, parameters minimizing noise in the Catphan showed noise reduction up to 20% in organs using Qr40. Across all reconstruction settings translated to patients for Br38 and Qr40, the maximum difference in mean HU value was 4.6. The maximum difference in noise (standard deviation) was 21%.

Conclusion: Reconstruction spatial resolution and contrast was impacted only by selected kernel. Voxel noise, however, depended on all parameters, with kernel being the most influential. The noise trends seen in the Catphan were found to translate to clinical anatomy.

ePosters

Keywords

Simulation, Reconstruction, Image Analysis

Taxonomy

Not Applicable / None Entered.

Contact Email