Evaluation of strategies to reduce radiation dose in perfusion CT imaging using a reproducible biologic phantom.

Abstract

The purpose of this study is to evaluate radiation dose reduction strategies in perfusion CT by using a biologic phantom. A formalin-preserved porcine liver was submerged in a 32-cm-wide acrylic phantom filled with water. The portal vein was connected to a continuous flow pump. The phantom was scanned with a perfusion CT protocol using 80 kVp and 400 mAs, every 1 second, for 50 seconds. This was repeated using 100 and 20 mAs. It was also repeated again using 400 mAs to assess reproducibility. A sparser scan frequency was simulated retrospectively. Blood flow was determined for each dataset using the maximum slope and deconvolution methods. Measurements of the mean blood flow values in identical regions of interest had a percent difference of 7% for repeated perfusion CT protocols using the same settings regardless of perfusion model used. The 100 mAs scans agreed with 400 mAs scans, with percent differences of 21% and 31% for the maximum slope and deconvolution methods, respectively. At a simulated frequency of one scan every 4 seconds, blood flow values differed up to 17% and 60% from the reference scan for the maximum slope and deconvolution methods, respectively. At 20 mAs and one scan every 1 second, or 400 mAs and a simulated frequency of one scan every 8 seconds, both computation methods failed to provide accurate blood flow estimates. The biologic phantom showed reproducible measurements that can help in optimizing perfusion CT protocols by determining both the acquisition parameters that affect radiation dose and the accuracy of estimates from different perfusion models.