Consistency of spectral results in cardiac dual-source photon-counting CT

Abstract

We evaluate stability of spectral results at different heart rates, acquisition modes, and cardiac phases in first-generation clinical dual-source photon-counting CT (PCCT). A cardiac motion simulator with a coronary stenosis mimicking a 50% eccentric calcium plaque was scanned at five different heart rates (0, 60-100 bpm) with the three available cardiac scan modes (high pitch prospectively ECG-triggered spiral, prospectively ECG-triggered axial, retrospectively ECG-gated spiral). Subsequently, full width half max (FWHM) of the stenosis, Dice score (DSC) for the stenosed region, and eccentricity of the non-stenosed region were calculated for virtual monoenergetic images (VMI) at 50, 70, and 150 keV and iodine density maps at both diastole and systole. FWHM averaged differences of - 0.20, - 0.28, and - 0.15 mm relative to static FWHM at VMI 150 keV across acquisition parameters for high pitch prospectively ECG-triggered spiral, prospectively ECG-triggered axial, and retrospectively ECG-gated spiral scans, respectively. Additionally, there was no effect of heart rate and acquisition mode on FWHM at diastole (p-values < 0.001). DSC demonstrated similarity among parameters with standard deviations of 0.08, 0.09, 0.11, and 0.08 for VMI 50, 70, and 150 keV, and iodine density maps, respectively, with insignificant differences at diastole (p-values < 0.01). Similarly, eccentricity illustrated small differences across heart rate and acquisition mode for each spectral result. Consistency of spectral results at different heart rates and acquisition modes for different cardiac phase demonstrates the added benefit of spectral results from PCCT to dual-source CT to further increase confidence in quantification and advance cardiovascular diagnostics.

Figure 1

Experimental set up. A cardiac motion phantom was scanned with a dual-source photon-counting CT (A). The phantom included rods mimicking coronary stenoses of different materials (B) and extents (C) represented in a cross section through the center of the rod. Orange regions correspond to the stenosis while the yellow region corresponds to outer vessel walls.

Figure 2

Adaptive thresholding of stenosis on VMI 150 keV images at varying cardiac phases and heart rates with spiral mode. Stenoses at different phases and heart rates were isolated with adaptive thresholding (purple) and demonstrated similar shape and size in comparison to the static scan (0 bpm).

Figure 3

Shape evaluation of shape for non-stenosed region of the rod with spiral mode. Ellipse (red) were fitted to rods isolated from adaptive thresholding. At different phases and heart rates, the shape of the rod remained the same with some distortion at higher heart rates.

Figure 4

Consistency of full width half max (FWHM) for line profile of stenosed region with various parameters. FWHM varied with heart rate at systole for VMI 50, 70 keV, and iodine density maps but not vary with heart rate or acquisition mode at diastole for all four spectral results. Differences from the static measurements were small at a maximum of 13%.

Figure 5

Stability of Dice score of the stenosed region for different spectral results. Dice score demonstrated different primarily associated with heart rate at systole. Clear variation in Dice score at 80 bpm was present, particularly for VMI 70 and 150 keV.

Figure 6

Eccentricity of the non-stenosed region of the rod at different spectral results for different acquisition mode, phase, and heart rate. Eccentricity was not affected by acquisition mode at both systole and diastole while was affected by heart rate, especially for sequence 35% at VMI 50 keV, spiral 35% at VMI 50 keV, and sequence 70% at VMI 70 keV and iodine density maps.