Deep learning reconstruction of triple-source CT data with sparse view and truncation

Cardiovascular disease remains a leading cause of global mortality, driving innovation in cardiac imaging. Interior tomography, which limits x-ray beams to a specific region of interest (ROI), significantly reduces radiation exposure, making it ideal for cardiac applications. A Triple-Source CT (TSCT) capable of performing interior tomography of the heart using three simultaneous imaging chains shows promise for rapid, low-dose cardiac imaging. However, conventional analytic reconstruction algorithms struggle with severe artifacts due to data truncation and sparse-view sampling. In this paper, we propose a novel interpolation-based dual-task network for deep reconstruction of cardiac CT images in TSCT. Our approach employs two sub-networks that address sparse-view and truncation artifacts separately, leveraging distinct ground truths to constrain each sub-network during training. This separation simplifies training and enables more focused and effective artifact removal. Additionally, we utilize two interpolation methods to complete the sinogram as a prior input, enhancing reconstruction accuracy. Experimental results from imaging a porcine heart using 84 views and a 31.3% truncation ratio demonstrate that our method effectively suppresses both artifact types while preserving image details. Compared to conventional FBP, our approach achieves improvements of 40% in RMSE and 6% in SSIM. This reconstruction method shows potential for cardiac imaging in TSCT systems.