One-step iterative reconstruction approach based on eigentissue decomposition for spectral photon-counting computed tomography

Abstract

Purpose: We propose a one-step tissue characterization method for spectral photon-counting computed tomography (SPCCT) using eigentissue decomposition (ETD), tailored for highly accurate human tissue characterization in radiotherapy. Methods:The approach combines a Poisson likelihood, a spatial prior, and a quantitative prior constraining eigentissue fractions based on expected values for tabulated tissues. There are two regularization parameters: α for the quantitative prior, and β for the spatial prior. The approach is validated in a realistic simulation environment for SPCCT. The impact of α and β is evaluated on a virtual phantom. The framework is tested on a virtual patient and compared with two sinogrambased two-step methods [using respectively filtered backprojection (FBP) and an iterative method for the second step] and a post-reconstruction approach with the same quantitative prior. All methods use ETD.Results: Optimal performance with respect to bias or RMSE is achieved with different combinations of α and β on the cylindrical phantom. Evaluated in tissues of the virtual patient, the one-step framework outperforms two-step and post-reconstruction approaches to quantify proton-stopping power (SPR). The mean absolute bias on the SPR is 0.6% (two-step FBP), 0.6% (two-step iterative), 0.6% (post-reconstruction), and 0.2% (one-step optimized for low bias). Following the same order, the RMSE on the SPR is 13.3%, 2.5%, 3.2%, and 1.5%.Conclusions: Accurate and precise characterization with ETD can be achieved with noisy SPCCT data without the need to rely on post-reconstruction methods. The one-step framework is more accurate and precise than two-step methods for human tissue characterization.

Dual- and multi-energy CT for particle stopping-power estimation: current state, challenges and potential

“…One-step methods resolve the image by formulating the problem as a large-scale optimization to extract image properties directly from spectral sinogram data. A few of those methods were proposed over the last decade, first for DECT (Cai et al 2013, Long and Fessler 2014, Jolivet et al 2021, later for spectral PCCT focused on K-edge imaging (Foygel Barber et al 2016, Weidinger et al 2016, and recently for spectral PCCT in the context of radiotherapy (Simard and Bouchard 2022). Such approaches are currently the most promising for simultaneously correcting major reconstruction artefacts (e.g.…”

“…With experimental data of tissue substitutes, they showed that the RMSE on ρ s decreases by a factor of 1.9 using the Bayesian technique. Another study by Simard et al (2022) demonstrated the robustness of the Bayesian approach with photon-counting CT (4 energy bins) using experimental measurements with tissue substitutes and 2 contrast agents.…”

“…Because each energy channel has a distinctive mean energy, the spectral attenuation information naturally discriminates materials by providing a more detailed behavior of the attenuation coefficient of a medium with respect to energy. Several studies have shown the potential to discriminate multiple contrast agents at once with MECT (Schlomka et al 2008, Cormode et al 2010, 2017, Si-Mohamed et al 2019, Dunning et al 2020, Simard et al 2022. The strategy is to attribute contrast agents to an equal number of basis materials and keep the remaining basis materials to describe human tissues without the agent.…”

Image quality improvement of a one-step spectral CT reconstruction on a prototype photon-counting scanner

“…However, these techniques are computationally more demanding and require an efficient optimization algorithm. Several algorithms have been investigated for one-step reconstruction: nonlinear conjugate gradient (Cai et al 2013, Simard andBouchard 2022), separable quadratic surrogates (SQS) (Long and Fessler 2014, Weidinger et al 2016, Mechlem et al 2017, Tilley et al 2019, Lee et al 2022, Liu et al 2022 or algorithms based on proximal operators such as ADMM (Jolivet et al 2020, Schmidt et al 2022, Chambolle-Pock (Barber et al 2016, Chen et al 2021 or VMILa (Tairi et al 2020). A previous comparative study (Mory et al 2018) has demonstrated that SQS combined with Nesterov's momentum is an efficient algorithm.…”

“…This study aimed at demonstrating the image quality improvement of one-step approaches over a prereconstruction MD method, called two-step in the following. Existing one-step algorithms have been compared to two-step methods (Mechlem et al 2017, Simard andBouchard 2022) but, to our knowledge, no study has rigorously compared the image quality of one-step and two-step material decomposition strategies. In this study, a one-step algorithm and a two-step algorithm were selected from the literature and implemented.…”