High-fidelity image deconvolution for low-dose cerebral perfusion CT imaging via low-rank and total variation regularizations

Cerebral perfusion computed tomography (PCT) provides a comprehensive and accurate noninvasive survey of the site of arterial occlusion by producing hemodynamic parameter maps (HPMs) in a qualitative and quantitative way. An HPM can be generally yielded through singular value decomposition (SVD)-based deconvolution approaches. However, due to their sequential scan protocol of PCT imaging, SVD-based deconvolution approaches are usually sensitive to noise, especially in low-dose cases. To obtain a high-fidelity HPM for low-dose PCT, in this study, we propose a high-fidelity image-domain deconvolution method that utilizes low-rank and total-variation (LR-TV) constraints. Specifically, the LR-TV constraints model both the spatio-temporal structure information and the low-rank characteristics present in the PCT data to mitigate the oscillations from noise. Subsequently, a modified Split-Bregman method is introduced to optimize the associated objective function. Both digital phantom and clinical patient data experiments are conducted to validate and evaluate the performance of the proposed LR-TV method. The experimental results demonstrate that the proposed LR-TV method can outperform the existing deconvolution approaches in high-fidelity HPM estimation.