摘要
Computedtomography(CT)playsanimportantroleinthefieldofmodernmedicalimaging.Reducingradiationexposuredosewithoutsignificantlydecreasingimage'squalityisalwaysacrucialissue.Inspiredbytheoutstandingperformanceoftotalvariation(TV)techniqueinCTimagereconstruction,aTVregularizationbasedBayesian-MAP(MAP-TV)isproposedtoreconstructthecaseofsparseviewprojectionandlimitedanglerangeimaging.Thismethodcansuppressthestreakartifactsandgeometricaldeformationwhilepreservingimageedges.Weusedorderedsubset(OS)techniquetoacceleratethereconstructionspeed.NumericalresultsshowthatMAP-TVisabletoreconstructaphantomwithbettervisualperformanceandquantitativeevaluationthanclassicalFBP,MLEMandquadratepriortoMAPalgorithms.Theproposedalgorithmcanbegeneralizedtocone-beamCTimagereconstruction.
出版日期
2017年02月12日(中国Betway体育网页登陆平台首次上网日期,不代表论文的发表时间)