MRIitesityormalizatio
Itesityormalizatioofmulti-chaelMRIimagesusigthemethodproposedbyNyuletal.2000.Itheorigialpaper,theauthorssuggestamethodwhereasetofstadardhistogramladmarksarelearedfromasetofMRIimages.Theseladmarksaretheusedtoequalizethehistogramsoftheimagestoormalize.Ibothlearigadtrasformatio,thehistogramsareusedtofidtheitesityladmarks.Iourimplmetatio,theladmarksarecomputedbasedothetotalrageofitesitiesisteadofthehistograms.
Howitworks:Theormalizatioiscarriedoutitwosteps:
Learigtheladmarkparameters:Fromasetoftraiigimages,theladmarkparametersarelearedusigthefuctiolear_itesity_parameters.Itesityparametersì_miadi_maxhavetobesetbytheuser.Thesetwovaluesestablishthemiimumadmaximumitesitiesofthestadarditesityscale.
methodT='splie';%ormethodT='liear';trai_im_path{1}='/path/to/images/1/t1.ii';trai_im_path{2}='/path/to/images/2/t1.ii';...trai_im_path{}='/path/to/images//t1.ii';i_mi=mi_itesity;i_max=max_itesity;%leartheparametersm_k=lear_itesity_ladmarks(trai_im_path,i_mi,i_max);Theoutputstructm_kcotaisthestadardladmarkslearedfromtheiputimages.Theseladmarksrefertothemiimumitesity,thesigalitesitydeciles{d10,...,d90},adthemaximumitesityofiterest.
ApplythetrasformatiofuctiotoeachoftheimagestoormalizeTheoutputstructm_kisusedtomaptheitesitiesofeachoftheiputimageswithrespecttothestadardscale.Theorigialpaperimplemetsafuctiothatmapsliearlytheiputitesitiesitothestadardhistogram.However,theauthorssuggestthatothermappigfuctioscabealsoused.Here,iputitesitiesaremappedusigaspliefuctio.
iput_image='/path/to/iput/image'out_ame='/path/to/iput/image/ormalized_sca'apply_itesity_trasformatio(iput_image,out_ame,m_k,methodT);Notes:Iputimageshavetobeskull-strippedforoptimalresults.Ifimagesareotskull-strippedbutbackgrouditesityis<0.05,themethodshouldalsowork.Withbackgrouditesitieshigherthathisthresholdtheladmarksmaybealteredisomeuexpectedwayduetoremaiigskull,fatoreyes.
Thecurretmethodusestheifti_toolsrepositoryavailablehere.AddittoyourMatlabpathoriitializetheicludedsubmoduleaftercloigtheprojectas:
gitsubmoduleiitgitsubmoduleupdateCredits:SergiValverde/NeuroImageComputigGroup.VisioadRoboticsIsitituteVICOROB(UiversityofGiroa)
评论