Per kinetic analysis method all the ICCs, R2, and slopes obtained when applying dynamic contouring thresholds on threshold derived VOIs are plotted. Nonsmoothed (left columns) and smoothed (right columns) data were separated. In most cases the nonsmoothed data have a wider distribution, however, the nonsmoothed data showed better combinations of ICC, R2, and slopes. The Patlak method showed no positive ICCs for any combination of thresholds.
CT perfusion calculated using the slope method plotted against [15O]H2O PET. Dynamic contouring thresholds of −200 HU (lower) and +500 HU (upper) were applied on the dynamic CT data. ICC: 0.83; R2: 0.90; and Y: 0.70 (LOI: Line of interest).
A Bland-Altman plot obtained using the data points from the correlation between the CTP (Slope method) (dynamic contouring thresholds: −200 and +500 HU) and [15O]H2O PET (Fig. 2 ).
This graph shows the effect of dynamic lower border thresholds. Tumor movement causes two large variabilities in tumor enhancement when no threshold is applied (1). After dynamic lower border thresholds were applied, −300 (2) and −200 (3), the tumor CT_TAC improved. The use of dynamic lower threshold of −300 showed the best agreement of CT derived perfusion values with the use [15O]H2O PET.
A gamma curve fitted through the arterial (a) and tumor CT_TAC (b). The Slope and Mullani-Gould method used the gamma curve to calculate perfusion; STM fitted through the tumor CT_TAC (c); Patlak Plot plotted using the tumor CT_TAC (d) (Int arterial: Integral of arterial TAC).
Article metrics loading...
Full text loading...