?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Learning+to+Address+Intra-segment+Misclassification+in+Retinal+Imaging&rft.creator=Zhou%2C+Y&rft.creator=Xu%2C+M&rft.creator=Hu%2C+Y&rft.creator=Lin%2C+H&rft.creator=Jacob%2C+J&rft.creator=Keane%2C+PA&rft.creator=Alexander%2C+DC&rft.description=Accurate+multi-class+segmentation+is+a+long-standing+challenge+in+medical+imaging%2C+especially+in+scenarios+where+classes+share+strong+similarity.+Segmenting+retinal+blood+vessels+in+retinal+photographs+is+one+such+scenario%2C+in+which+arteries+and+veins+need+to+be+identified+and+differentiated+from+each+other+and+from+the+background.+Intra-segment+misclassification%2C+i.e.+veins+classified+as+arteries+or+vice+versa%2C+frequently+occurs+when+arteries+and+veins+intersect%2C+whereas+in+binary+retinal+vessel+segmentation%2C+error+rates+are+much+lower.+We+thus+propose+a+new+approach+that+decomposes+multi-class+segmentation+into+multiple+binary%2C+followed+by+a+binary-to-multi-class+fusion+network.+The+network+merges+representations+of+artery%2C+vein%2C+and+multi-class+feature+maps%2C+each+of+which+are+supervised+by+expert+vessel+annotation+in+adversarial+training.+A+skip-connection+based+merging+process+explicitly+maintains+class-specific+gradients+to+avoid+gradient+vanishing+in+deep+layers%2C+to+favor+the+discriminative+features.+The+results+show+that%2C+our+model+respectively+improves+F1-score+by+4.4%25%2C+5.1%25%2C+and+4.2%25+compared+with+three+state-of-the-art+deep+learning+based+methods+on+DRIVE-AV%2C+LES-AV%2C+and+HRF-AV+data+sets.+Code%3A+https%3A%2F%2Fgithub.com%2Frmaphoh%2FLearning-AVSegmentation&rft.subject=Multi-class+Segmentation%2C+Intra-segment+Misclassification%2C++Retinal+Vessel%2C+Binary-to-multi-class+Fusion+Network&rft.publisher=Springer&rft.date=2021-09-21&rft.type=Proceedings+paper&rft.language=eng&rft.source=+++++In%3A++Medical+Image+Computing+and+Computer+Assisted+Intervention+%E2%80%93+MICCAI+2021.++(pp.+pp.+482-492).++Springer%3A+Cham%2C+Switzerland.+(2021)+++++&rft.format=text&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10137248%2F1%2FMICCAI2021_camera_ready.pdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10137248%2F&rft.rights=open