000134894 001__ 134894
000134894 005__ 20240229105047.0
000134894 0247_ $$2doi$$a10.1007/s11548-018-1772-0
000134894 0247_ $$2pmid$$apmid:29704196
000134894 0247_ $$2ISSN$$a1861-6410
000134894 0247_ $$2ISSN$$a1861-6429
000134894 0247_ $$2altmetric$$aaltmetric:41853440
000134894 037__ $$aDKFZ-2018-00682
000134894 041__ $$aeng
000134894 082__ $$a610
000134894 1001_ $$00000-0002-7094-4926$$aRoss, Tobias$$b0$$eFirst author
000134894 245__ $$aExploiting the potential of unlabeled endoscopic video data with self-supervised learning.
000134894 260__ $$aBerlin$$bSpringer$$c2018
000134894 3367_ $$2DRIVER$$aarticle
000134894 3367_ $$2DataCite$$aOutput Types/Journal article
000134894 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1695987787_24399
000134894 3367_ $$2BibTeX$$aARTICLE
000134894 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000134894 3367_ $$00$$2EndNote$$aJournal Article
000134894 500__ $$aE132 entspricht E230
000134894 520__ $$aSurgical data science is a new research field that aims to observe all aspects of the patient treatment process in order to provide the right assistance at the right time. Due to the breakthrough successes of deep learning-based solutions for automatic image annotation, the availability of reference annotations for algorithm training is becoming a major bottleneck in the field. The purpose of this paper was to investigate the concept of self-supervised learning to address this issue.Our approach is guided by the hypothesis that unlabeled video data can be used to learn a representation of the target domain that boosts the performance of state-of-the-art machine learning algorithms when used for pre-training. Core of the method is an auxiliary task based on raw endoscopic video data of the target domain that is used to initialize the convolutional neural network (CNN) for the target task. In this paper, we propose the re-colorization of medical images with a conditional generative adversarial network (cGAN)-based architecture as auxiliary task. A variant of the method involves a second pre-training step based on labeled data for the target task from a related domain. We validate both variants using medical instrument segmentation as target task.The proposed approach can be used to radically reduce the manual annotation effort involved in training CNNs. Compared to the baseline approach of generating annotated data from scratch, our method decreases exploratively the number of labeled images by up to 75% without sacrificing performance. Our method also outperforms alternative methods for CNN pre-training, such as pre-training on publicly available non-medical (COCO) or medical data (MICCAI EndoVis2017 challenge) using the target task (in this instance: segmentation).As it makes efficient use of available (non-)public and (un-)labeled data, the approach has the potential to become a valuable tool for CNN (pre-)training.
000134894 536__ $$0G:(DE-HGF)POF3-315$$a315 - Imaging and radiooncology (POF3-315)$$cPOF3-315$$fPOF III$$x0
000134894 588__ $$aDataset connected to CrossRef, PubMed,
000134894 7001_ $$0P:(DE-HGF)0$$aZimmerer, David$$b1
000134894 7001_ $$0P:(DE-He78)59550e3c9ae4b46b714843863e0db8d9$$aVemuri, Anant$$b2$$udkfz
000134894 7001_ $$0P:(DE-He78)7ea9af59d03ec7deb982a0e0562358fa$$aIsensee, Fabian$$b3$$udkfz
000134894 7001_ $$0P:(DE-He78)1042737c83ba70ec508bdd99f0096864$$aWiesenfarth, Manuel$$b4$$udkfz
000134894 7001_ $$aBodenstedt, Sebastian$$b5
000134894 7001_ $$aBoth, Fabian$$b6
000134894 7001_ $$aKessler, Philip$$b7
000134894 7001_ $$aWagner, Martin$$b8
000134894 7001_ $$aMüller, Beat$$b9
000134894 7001_ $$aKenngott, Hannes$$b10
000134894 7001_ $$aSpeidel, Stefanie$$b11
000134894 7001_ $$0P:(DE-He78)bb6a7a70f976eb8df1769944bf913596$$aKopp-Schneider, Annette$$b12$$udkfz
000134894 7001_ $$0P:(DE-He78)33c74005e1ce56f7025c4f6be15321b3$$aMaier-Hein, Klaus$$b13$$udkfz
000134894 7001_ $$0P:(DE-He78)26a1176cd8450660333a012075050072$$aMaier-Hein, Lena$$b14$$eLast author$$udkfz
000134894 773__ $$0PERI:(DE-600)2235881-X$$a10.1007/s11548-018-1772-0$$gVol. 13, no. 6, p. 925 - 933$$n6$$p925 - 933$$tInternational journal of computer assisted radiology and surgery$$v13$$x1861-6429$$y2018
000134894 909CO $$ooai:inrepo02.dkfz.de:134894$$pVDB
000134894 9101_ $$0I:(DE-588b)2036810-0$$60000-0002-7094-4926$$aDeutsches Krebsforschungszentrum$$b0$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-HGF)0$$aDeutsches Krebsforschungszentrum$$b1$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)59550e3c9ae4b46b714843863e0db8d9$$aDeutsches Krebsforschungszentrum$$b2$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)7ea9af59d03ec7deb982a0e0562358fa$$aDeutsches Krebsforschungszentrum$$b3$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)1042737c83ba70ec508bdd99f0096864$$aDeutsches Krebsforschungszentrum$$b4$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)bb6a7a70f976eb8df1769944bf913596$$aDeutsches Krebsforschungszentrum$$b12$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)33c74005e1ce56f7025c4f6be15321b3$$aDeutsches Krebsforschungszentrum$$b13$$kDKFZ
000134894 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)26a1176cd8450660333a012075050072$$aDeutsches Krebsforschungszentrum$$b14$$kDKFZ
000134894 9131_ $$0G:(DE-HGF)POF3-315$$1G:(DE-HGF)POF3-310$$2G:(DE-HGF)POF3-300$$3G:(DE-HGF)POF3$$4G:(DE-HGF)POF$$aDE-HGF$$bGesundheit$$lKrebsforschung$$vImaging and radiooncology$$x0
000134894 9141_ $$y2018
000134894 915__ $$0StatID:(DE-HGF)0200$$2StatID$$aDBCoverage$$bSCOPUS
000134894 915__ $$0StatID:(DE-HGF)0300$$2StatID$$aDBCoverage$$bMedline
000134894 915__ $$0StatID:(DE-HGF)0100$$2StatID$$aJCR$$bINT J COMPUT ASS RAD : 2015
000134894 915__ $$0StatID:(DE-HGF)0199$$2StatID$$aDBCoverage$$bThomson Reuters Master Journal List
000134894 915__ $$0StatID:(DE-HGF)0111$$2StatID$$aWoS$$bScience Citation Index Expanded
000134894 915__ $$0StatID:(DE-HGF)0150$$2StatID$$aDBCoverage$$bWeb of Science Core Collection
000134894 915__ $$0StatID:(DE-HGF)1110$$2StatID$$aDBCoverage$$bCurrent Contents - Clinical Medicine
000134894 915__ $$0StatID:(DE-HGF)9900$$2StatID$$aIF < 5
000134894 9201_ $$0I:(DE-He78)E130-20160331$$kE130$$lE130 Intelligente Medizinische Systeme$$x0
000134894 9201_ $$0I:(DE-He78)E132-20160331$$kE132$$lMedizinische Bildverarbeitung$$x1
000134894 9201_ $$0I:(DE-He78)E230-20160331$$kE230$$lE230 Medizinische Bildverarbeitung$$x2
000134894 9201_ $$0I:(DE-He78)C060-20160331$$kC060$$lC060 Biostatistik$$x3
000134894 980__ $$ajournal
000134894 980__ $$aVDB
000134894 980__ $$aI:(DE-He78)E130-20160331
000134894 980__ $$aI:(DE-He78)E132-20160331
000134894 980__ $$aI:(DE-He78)E230-20160331
000134894 980__ $$aI:(DE-He78)C060-20160331
000134894 980__ $$aUNRESTRICTED