Spectral CT Image Restoration via an Average Image-Induced Nonlocal Means Filter PROJECT TITLE :Spectral CT Image Restoration via an Average Image-Induced Nonlocal Means FilterABSTRACT:Goal: Spectral computed tomography (SCT) images reconstructed by an analytical approach typically suffer from a poor signal-to-noise ratio and strong streak artifacts when sufficient photon counts don't seem to be accessible in SCT imaging. In reducing noise-induced artifacts in SCT pictures, during this study, we propose a mean image-induced nonlocal means that (aviNLM) filter for every energy-specific image restoration. Methods: The current aviNLM algorithm exploits redundant data in the whole energy domain. Specifically, the proposed aviNLM algorithm yields the restored results by performing a nonlocal weighted average operation on the noisy energy-specific images with the nonlocal weight matrix between the target and previous pictures, in that the previous image is generated from all of the pictures reconstructed in every energy bin. Results: Qualitative and quantitative studies are conducted to guage the aviNLM filter by using the info of digital phantom, physical phantom, and clinical patient knowledge acquired from the energy-resolved and -integrated detectors, respectively. Experimental results show that the current aviNLM filter will achieve promising results for SCT image restoration in terms of noise-induced artifact suppression, cross profile, and contrast-to-noise ratio and material decomposition assessment. Conclusion and Significance: The present aviNLM algorithm has useful potential for radiation dose reduction by lowering the mAs in SCT imaging, and it may be helpful for another clinical applications, like in myocardial perfusion imaging and radiotherapy. Did you like this research project? To get this research project Guidelines, Training and Code... Click Here facebook twitter google+ linkedin stumble pinterest Quantile-Based Simulation Optimization With Inequality Constraints: Methodology and Applications Neural Data-Driven Musculoskeletal Modeling for Personalized Neurorehabilitation Technologies