Redundancy information-induced image reconstruction for low-dose myocardial perfusion computed tomography

OBJECTIVE: In the clinic, myocardial perfusion computed tomography (MPCT) imaging is commonly used to detect and assess myocardial ischemia quantitatively. However, repeated scanning on the myocardial region in the cine mode will increase the radiation dose for patients. With lowering radiation dose, the quality of images are degraded by noise induced artifact, which hampers the diagnostic accuracy. Therefore, in this paper, we propose a redundancy information induced iterative reconstruction framework for high quality MPCT images at the case of low dose. METHODS: MPCT images have redundant structural information within frames and highly similarity between adjacent frames. Inspired by the two properties, in this work we propose a penalized weighted least-squares (PWLS) model incorporating NLM and TV based hybrid constraints, which is referred to as PWLS-aviNLM-TV for simplicity. The proposed algorithm can effectively eliminate noise and artifacts by taking into account the similarity between adjacent frames and redundancy information within frames, which also can improve spatial resolution within frames and maintain temporal resolution. RESULTS: The experimental results on the 4D extended cardiac-torso (XCAT) phantom and preclinical porcine dataset demonstrates that the PWLS-aviNLM-TV algorithm obtains better performance in terms of noise reduction and artifacts suppression than the PWLS-TV and PWLSaviNLM algorithm. Moreover, the proposed algorithm can preserve the edges and detail information thereby efficiently differentiate ischemia from myocardium. CONCLUSIONS: The present redundancy information induced reconstruction algorithm can reconstruct high-quality images from low-dose MPCT for better clinical imaging diagnosis.