Investigation of HYPR image voxel noise vs 3D cubic filter kernel width for 5, 10, and 20 beat composite windows (measurements shown are from the 100 mA scan of Pig 1). Solid horizontal lines indicate the noise level measured in the composite image.
HYPR image voxel noise vs composite window size and tube current using a 7 voxel wide filter kernel (Pig 1). The dashed horizontal line indicates the noise measured in the 500 mA FBP images.
FBP versus HYPR image reconstructions at the same time point in a 100 mA scan (top row), a 50 mA scan (middle row), and a 25 mA scan (bottom row). A FBP reconstruction from the 500 mA scan is shown on the upper right for comparison. The field of view has been cropped around the cardiac anatomy (Pig 1).
Voxel noise in HYPR images (circles) and FBP images (squares) versus mA, expressed relative to the noise measured in 500 mA FBP images. Bars represent the standard deviation in the individual noise ratio measurements.
HYPR and FBP time-attenuation curves measured from a large, 15 216 voxel (2.3 cm3) volume of segment 1 in the 100 mA scan of Pig 4.
Average limits of agreement between HYPR and FBP TACs versus tube current.
HYPR and FBP time-attenuation curves measured from a small, 492 voxel (0.07 cm3) volume of segment 1 in the 50 mA scan of Pig 4.
RMSD of TAC values relative to curve fit, as function of VOI size, in segment 1 of the 50 mA scan of Pig 4. Dashed lines show the calculation of HYPR and FBP VOI sizes at a RMSD threshold value of 3 HU.
Smallest time-attenuation curve measurement volume for HYPR versus FBP images, as a function of tube current. Symbols represent the average over all segments and pigs. Bars represent the range of individual segment results, after averaging over all pigs.
Composite window sizes projected to achieve 500 mA FBP noise levels and selected values.
Article metrics loading...
Full text loading...