Adam Grant Chandler PhD, Presenter: Employee, General Electric Company

Delise H. Herron RN, Abstract Co-Author: Nothing to Disclose

Ella Anderson, Abstract Co-Author: Nothing to Disclose

Wei Wei, Abstract Co-Author: Nothing to Disclose

James C Yao, Abstract Co-Author: Nothing to Disclose

Chaan Ng MD, Abstract Co-Author: Nothing to Disclose

Previous work has shown that a sampling interval (SI) of 3s is the approximate temporal sampling upper limit for perfusion in the brain and in the pelvic cavity. As long as the SI is ~3s or less then the blood flow (BF) and blood volume (BV) values calculated by the CT Perfusion software should be accurate. The aim of this work was to determine if such a statement is true for perfusion parameters in abdominal tissues (namely, liver, spleen and kidney).

CTP data was acquired in ten patients with liver tumors on a 64 row CT (8x5mm thickness, 30s end-inspiration breathold cine acquisition, 50cc intravenous contrast). For each dataset, the original series (0.5s SI) was reformatted to SIs of 1.0, 2.0 and 3.0s. BF and BV parametric maps for the original and three reformatted series’ were computed using commercially available CT perfusion software (CT Perfusion 4, General Electric Healthcare). The same aortic input region was used for the original and reformatted series, as were the last pre-enhancement and first post-enhancement times. All other algorithmic parameters were kept the same. This was to ensure that the only changing variable was the SI. For each dataset, tumor, normal liver, spleen and kidney regions of interest (ROI’s) were defined on a slice-by-slice basis on the original 0.5s SI series. For each ROI and each reformatted series, BF and BV values were recorded and plotted against 0.5s SI values for that ROI. Linear regression slope coefficients and R2 values were was used to compute the correlation and fit.

In total, 228 ROI were analyzed (108 tumor; 46 normal liver; 53 spleen; 21 kidney). BF showed relatively stable behavior as SI was increased from 0.5s to 3s, with slope coefficients close to unity, in the range 0.974 to 1.022, and with R2 values in the range 0.982 to 0.997. BV showed similar behavior with slope coefficients in the range 0.973 to 1.013, and R2 values in the range 0.972 to 0.976.

SI as large as 3s permits relatively accurate estimates of BV and BF in abdominal tissue compared to those obtained with 0.5s intervals. By increasing the SI in such studies, the dose delivered in a liver CTP protocol can be reduced.

The ability to reduce temporal sampling in CT perfusion studies without compromising estimates of perfusion parameters would enable reductions in radiation exposure.

Chandler, A, Herron, D, Anderson, E, Wei, W, Yao, J, Ng, C, The Effects of Temporal Sub-Sampling on Estimates of CT Perfusion Parameters in Abdominal Tissues.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8009334.html