Room: Room 202
Purpose: To develop a multi-energy iterative CT reconstruction framework for a whole-body research photon-counting detector based computed tomography (PCD-CT) system to achieve low noise, high resolution energy bin image reconstruction with high spatial and spectral fidelity.
Methods: We propose a spectral prior image constrained compressed sensing CT reconstruction approach that incorporates an edge-preserving non-local means filtered prior (NLM-SPICCS) for multi-energy CT data from a research whole-body PCD-CT system. Imaging experiments were performed to evaluate NLM-SPICCS: (i) Anthropomorphic head phantom in macro mode (32x0.5 mm collimation, standard resolution) at 140kV, 66 mAs with [25,75] keV energy thresholds, (ii) thin wire phantom scan at 140 kV, 110 mAs using ultra-high resolution (UHR, 32x0.25 mm collimation) mode to measure spatial resolution, and (iii) a pig head scanned using UHR mode at 120kV, 202 mAs and [35,52] keV energy thresholds. The prior image for NLM-SPICCS was obtained using conventional filtered-back projection (FBP) technique, followed by NLM filter operation. Mean CT number and standard deviation within ROIs were measured from the bin images. Modulation transfer function (MTF) was measured for conventional FBP, conventional SPICCS and NLM-SPICCS.
Results: The head phantom bin reconstructions (25-75 keV) demonstrated a 39% noise reduction from NLM-SPICCS relative to conventional FBP. The NLM-SPICCS technique provided an additional 17% noise reduction compared to conventional SPICCS. Spatial resolution measurements showed a 10% MTF of 1.5mmâ?»Â¹ from FBP and 1.7mmâ?»Â¹ from conventional SPICCS and NLM-SPICCS. The pig head data showed a noise reduction of 80% in 35-52 keV bin image from NLM-SPICCS relative to conventional FBP, while the CT numbers in the representative ROIs were consistent.
Conclusion: Additional noise reduction could be achieved using the SPICCS framework when an edge-preserving image-domain denoising is performed on the prior. This synergistic approach is beneficial for reconstructing bin images from both standard and ultra-high resolution PCD-CT.
Funding Support, Disclosures, and Conflict of Interest: This research was supported in part by NIH Grant, R01-EB016966 and C06-RR018898. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The research photon-counting CT system described herein is not commercially available.