Purpose: This study applies the one-step inversion material decomposition algorithm, cOSSCIR, to Dual kV spectral CT and validates our approach on digital and experimental phantom data.
Methods: The cOSSCIR algorithm is a one-step inversion approach in which the basis images are directly optimized from the spectral projection measurements. cOSSCIR demonstrated reduced artifacts caused by beam hardening and photon starvation for photon-counting CT. The cOSSCIR approach does not require spectral measurements to be registered and thus may provide additional benefits for clinical Dual kV CT systems. By modifying the spectral response and accounting for the detector gain, an integrating detector model was developed and tested on a simple digital phantom. A subsequent inverse crime study was used to validate the approach. A preliminary experimental study was also performed using a physical phantom composed of cylindrical rods of PMMA, Polystyrene, and Teflon. Projection data was acquired at 80 and 130 kV over 500 equally spaced view angles. The system consists of a micro-focal x-ray source (Hamamatsu 9181-02) and a high resolution flat panel detector (Varian PaxScan 2520 DX). The Aluminum and PMMA basis images were then reconstructed using cOSSCIR. For each basis image, the mean value in each material ROI was compared with their predicted values.
Results: Convergence to the digital phantom suggests that our integrating detector model is consistent with the mathematical framework developed for cOSSCIR. Furthermore, the experimental results demonstrate good recovery of the materials present within the physical phantom. The percent error in the PMMA basis image was 1.9% 3.3%, 4.3% in the polystyrene, PMMA, and Teflon ROIs. Similarly, in the Aluminum basis image the percent error was 7.5%, and 38.4% in the Polystyrene and Teflon ROIs.
Conclusion: By incorporating an integrating detector model into the cOSSICR framework we demonstrate direct inversion material decomposition for Dual kV data.
Funding Support, Disclosures, and Conflict of Interest: NIH R01EB023968