Franco Rupcich BS, BEng, Presenter: Nothing to Disclose

Andreu Badal-Soler, Abstract Co-Author: Nothing to Disclose

Iacovos Kyprianou PhD, Abstract Co-Author: Nothing to Disclose

Taly Gilat Schmidt PhD, Abstract Co-Author: Nothing to Disclose

To quantify breast dose and task-based image quality of clinically available methods for reducing breast dose during CT. Two task-based metrics were compared against pixel SNR to determine the benefits of task-based evaluation.

We simulated chest CT acquisitions of a 26 year old female phantom for a 120 kV reference CT protocol and four protocols suggested to reduce breast dose: 120 kV angular tube-current modulation (TCM); 120 kV partial scan (232°); 80 kV; 80 kV partial scan. A 2mm diameter, low-contrast (3 mg/mL) iodine detection task was located in the locally uniform heart. Ray-tracing was used to acquire projections. Breast dose was estimated with a validated dose database and the number of photons emitted during ray-tracings selected to achieve equivalent breast dose among protocols. Images were reconstructed with filtered back-projection. For each protocol we calculated the pixel SNR, Hotelling observer SNR (HOSNR), search-based SNR, and noise standard deviation (SD).  

For the 120 kV TCM scan, the SNR relative to the reference protocol at equivalent breast dose was 0.99, 0.98, and 0.87 for the pixel SNR, HOSNR, and search-based SNR, respectively. Also respectively and at equivalent breast dose, relative SNRs were 1.16, 1.12, and 1.16 for the 120 kV partial scan; 2.63, 2.56, and 4.06 for the 80 kV scan; 3.16, 2.9, and 4.96 for the 80 kV partial scan. At equivalent HOSNR to the reference protocol, 80 kV and 80 kV partial scans reduced breast dose by 85% and 88%, respectively, with 250% and 229% noise SD increase. The 120 kV partial-scan reduced breast dose by 21% with 5% noise SD increase. The 120 kV TCM scan increased breast dose by 4% with 2% SD decrease.

For this simulated task, 80 kV protocols provided the largest reduction (~85%) in breast dose while maintaining equivalent HOSNR to the reference protocol. However, the resulting ~200% noise SD increase may not be clinically acceptable, thus 80 kV may instead be beneficial for reducing contrast concentration. 120 kV partial-scan techniques may reduce dose while maintaining similar HOSNR and noise. Pixel SNR and HOSNR performed similarly, while search-based SNR demonstrated larger differences between protocols.

Studies suggest increased cancer risk for women from cardiac CT, partly due to breast dose. This work quantified breast dose and task-based image quality of available dose-reduction methods.

Rupcich, F, Badal-Soler, A, Kyprianou, I, Schmidt, T, Reducing Breast Dose during CT: A Unified Study of Dose and Task-based Image Quality.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12043859.html