%0 Journal Article %T Optimized Parallelization for Nonlocal Means Based Low Dose CT Image Processing %+ General Hospital Shenyang %+ Laboratory of Image Science and Technology [Nanjing] (LIST) %+ Centre de Recherche en Information Biomédicale sino-français (CRIBS) %+ Lab of Image Science and Technology – Key Laboratory of Computer Network and Information Integration (Southeast University), %A Zhang, Libo %A Yang, Benqiang %A Zhuang, Zhikun %A Hu, Yining %A Chen, Yang %A Luo, Limin %A Shu, Huazhong %Z 2013225089, National Basic Research Program of China %< avec comité de lecture %@ 1748-670X %J Computational and Mathematical Methods in Medicine %I Hindawi Publishing Corporation %V 2015 %P 790313 %8 2015 %D 2015 %R 10.1155/2015/790313 %M 26078781 %K Algorithms %K Artifacts %K Computational Biology %K Humans %K Models %K Statistical %K Phantoms %K Imaging %K Radiation Dosage %K Radiographic Image Interpretation %K Computer-Assisted %K Signal-To-Noise Ratio %K Tomography %K X-Ray Computed %Z Life Sciences [q-bio]/BioengineeringJournal articles %X Low dose CT (LDCT) images are often significantly degraded by severely increased mottled noise/artifacts, which can lead to lowered diagnostic accuracy in clinic. The nonlocal means (NLM) filtering can effectively remove mottled noise/artifacts by utilizing large-scale patch similarity information in LDCT images. But the NLM filtering application in LDCT imaging also requires high computation cost because intensive patch similarity calculation within a large searching window is often required to be used to include enough structure-similarity information for noise/artifact suppression. To improve its clinical feasibility, in this study we further optimize the parallelization of NLM filtering by avoiding the repeated computation with the row-wise intensity calculation and the symmetry weight calculation. The shared memory with fast I/O speed is also used in row-wise intensity calculation for the proposed method. Quantitative experiment demonstrates that significant acceleration can be achieved with respect to the traditional straight pixel-wise parallelization %G English %L hal-01326302 %U https://univ-rennes.hal.science/hal-01326302 %~ UNIV-RENNES1 %~ LTSI %~ CRIBS %~ STATS-UR1 %~ UR1-HAL %~ UR1-MATH-STIC %~ TEST-UNIV-RENNES %~ TEST-UR-CSS %~ UNIV-RENNES %~ UR1-MATH-NUM %~ UR1-BIO-SA