Understanding and Interpreting Machine Learning in Medical Image Computing Applications: First International Workshops, MLCN 2018, DLF 2018, and iMIMIC 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 16-20, 2018, Proceedings PDF
Preview Understanding and Interpreting Machine Learning in Medical Image Computing Applications: First International Workshops, MLCN 2018, DLF 2018, and iMIMIC 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 16-20, 2018, Proceedings
Danail Stoyanov · Zeike Taylor Seyed Mostafa Kia · Ipek Oguz Mauricio Reyes et al. (Eds.) Understanding and Interpreting 8 3 Machine Learning 0 1 1 in Medical Image Computing S C N Applications L First International Workshops MLCN 2018, DLF 2018, and iMIMIC 2018 Held in Conjunction with MICCAI 2018 Granada, Spain, September 16–20, 2018, Proceedings 123 Lecture Notes in Computer Science 11038 Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen Editorial Board David Hutchison Lancaster University, Lancaster, UK Takeo Kanade Carnegie Mellon University, Pittsburgh, PA, USA Josef Kittler University of Surrey, Guildford, UK Jon M. Kleinberg Cornell University, Ithaca, NY, USA Friedemann Mattern ETH Zurich, Zurich, Switzerland John C. Mitchell Stanford University, Stanford, CA, USA Moni Naor Weizmann Institute of Science, Rehovot, Israel C. Pandu Rangan Indian Institute of Technology Madras, Chennai, India Bernhard Steffen TU Dortmund University, Dortmund, Germany Demetri Terzopoulos University of California, Los Angeles, CA, USA Doug Tygar University of California, Berkeley, CA, USA Gerhard Weikum Max Planck Institute for Informatics, Saarbrücken, Germany More information about this series at http://www.springer.com/series/7412 Danail Stoyanov Zeike Taylor (cid:129) Seyed Mostafa Kia Ipek Oguz (cid:129) Mauricio Reyes et al. (Eds.) Understanding and Interpreting Machine Learning in Medical Image Computing Applications First International Workshops MLCN 2018, DLF 2018, and iMIMIC 2018 Held in Conjunction with MICCAI 2018 Granada, Spain, September 16-20, 2018 Proceedings 123 Editors DanailStoyanov IpekOguz University CollegeLondon Vanderbilt University London,UK Nashville, TN,USA ZeikeTaylor Mauricio Reyes University of Leeds University of Bern Leeds,UK Bern, Switzerland Seyed MostafaKia Radboud University Medical Center Nijmegen, TheNetherlands Additional WorkshopEditors seenext page ISSN 0302-9743 ISSN 1611-3349 (electronic) Lecture Notesin Computer Science ISBN 978-3-030-02627-1 ISBN978-3-030-02628-8 (eBook) https://doi.org/10.1007/978-3-030-02628-8 LibraryofCongressControlNumber:2018958518 LNCSSublibrary:SL6–ImageProcessing,ComputerVision,PatternRecognition,andGraphics ©SpringerNatureSwitzerlandAG2018 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartofthe material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodologynow knownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbookare believedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsortheeditors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictionalclaimsin publishedmapsandinstitutionalaffiliations. ThisSpringerimprintispublishedbytheregisteredcompanySpringerNatureSwitzerlandAG Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland Additional Workshop Editors Tutorial and Educational Chair Anne Martel University of Toronto Toronto, ON Canada Workshop and Challenge Co-chair Lena Maier-Hein German Cancer Research Center (DKFZ) Heidelberg Germany First International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2018 Andre F. Marquand Tommy Löfstedt Radboud University Umeå University Nijmegen Umeå The Netherlands Sweden Edouard Duchesnay NeuroSpin, CEA Saclay, Paris France First International Workshop on Deep Learning Fails, DLF 2018 Bennett Landman M. Jorge Cardoso Vanderbilt University King’s College London Nashville, TN London USA UK VI Additional Workshop Editors First International Workshop on Interpretability of Machine Intelligence in Medical Image Computing, iMIMIC 2018 Carlos A. Silva Raphael Meier University of Minho University of Bern Guimarães Bern Portugal Switzerland Sérgio Pereira University of Minho Guimarães Portugal MLCN 2018 Preface The first international workshop on Machine Learning in Clinical Neuroimaging (MLCN)washeldinconjunctionwithMICCAI2018,withaspecialfocusonspatially structureddataanalysis.Theworkshopaimedtobringtogethertop-notchresearchersin machinelearningandclinicalneurosciencetodiscussandhopefullybridgetheexisting gap in applied machine learning in clinical neuroscience. The main objective was to shed light on the opportunities and challenges in the structure-aware modeling in neuroimaging data in both encoding and decoding settings. For the keynote talks, leading researchers in the domain of spatial statistics, pattern recognition in neu- roimaging, and predictive clinical neuroscience, Prof. Christos Davatzikos (University of Pennsylvania), Dr. Gael Varoquaux (Inria), and Dr. George Langs (Medical UniversityofVienna),wereinvitedinordertoprovideacomprehensiveoverviewfrom theory to application in the field. The call for papers for the MLCN 2018 workshop was released on April 1, 2018, with the paper deadline set to July 25, 2018. The seven manuscripts received went through a double-blind review process by the MLCN scientific committee (Ehsan Adeli, Andre Altman, Luca Ambrogioni, Richard Dinga, Koen Haak, Christina Isa- koglou, Emanuele Olivetti, Pradeep Reddy Raamana, Kerstin Ritter, Sourena Soheili Nezhad, Thomas Wolfers, and Maryam Zabihi), and the top four papers with the best reviews were accepted for publication in the proceedings. The accepted contributions develop state-of-the-art machine learning methods such as spatio-temporal Gaussian processanalysis, stochasticvariational inference,anddeeplearningforapplications in Alzheimer’s disease diagnosis and multi-site neuroimaging data analysis. September 2018 Seyed Mostafa Kia Andre Marquand Edouard Duchesnay Tommy Löfstedt DLF 2018 Preface Deep learning methods have rapidly become omnipresent within the medical image computing and computer assisted intervention (MICCAI) community in recent years, thanks to their many attractive properties including state-of-the-art accuracy in many tasks in areas such as segmentation and classification. However, now that the initial excitement about these new techniques has ledtomany successful applicationswithin theMICCAIdomain,weneedtobegindevelopingabetterunderstandingtodemystify deep learning. To this end, we hosted a new workshop in conjunction with MICCAI 2018 (September 16–20, 2018, Granada, Spain) dedicated to understanding the “edges” of deeplearning:Whatareitscurrentlimitations?WhataresomeMICCAIproblemsthat are not well-suited for existing DL methods? What are some failures the community has encountered in DL? How can we better understand the “mysteries” we encounter, whether an algorithm works unexpectedly well or unexpectedly poorly? Where is the field going? etc. The workshop was held on September 16, 2018, in the Granada Exhibition and Conference Center. Submissions were solicited via a call for papers that was widely circulated. Some example ideas for possible contributions were suggested, but more importantly, we invited the MICCAI community to brainstorm about deep learning in the context of MICCAI-related fields. Wewerepleasedtoobservethecommunity respondedwell to this invitation and the submissions covered a wide range of topics. Each submission underwent a single-blind review by at least two members of the Program Committee, which consisted of researchers who actively contribute in the area. At the conclusion ofthereviewprocess,sixpapers wereaccepted.Theprogramwasfurtherenrichedby two invited speakers, Dr. Scott Acton and Dr. Leo Grady. We chose the (provocative on purpose) title of “Deep Learning Fails” for this workshop,tongue-in-cheek.Tobeclear,thegoalofthisworkshopwasnottodisparage deeplearningmethods.Assuch,wedidnotallowpapersthatindiscriminatelytrivialize deeplearning,suchasapapershowingnegativeresultsusingagenericnetworkmodel that has not been adapted or fine-tuned to address a specific problem. Rather, we encouraged submissions that are in the spirit of constructive criticism, with the aim of evaluating the strengths and weaknesses of deep learning, as well as identifying the main challenges in the current state-of-the-art and future directions. We would like to thank everyone who helped make this workshop happen: the authors who contributed their work, the Program Committee for their careful and thoughtful review, the invited speakers for sharing their expertise and insights, the attendees for their contribution to the discussion, and the MICCAI society for general support. September 2018 Ipek Oguz Bennett Landman Jorge M. Cardoso iMIMIC 2018 Preface ThefirsteditionoftheworkshoponInterpretabilityofMachineIntelligenceinMedical Image Computing (iMIMIC)1 was held on September 16, 2018, as a half-day satellite event of the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), in Granada, Spain. With its first edition, thisworkshopaimedatintroducingthechallengesandopportunitiesofinterpretability of machine learning systems in the context of MICCAI, as well as understanding the currentstateoftheartinthetopicandpromotingitasacrucialareaforfurtherresearch. Theworkshopprogramcomprisedoforalpresentationsoftheacceptedworksandtwo keynotes provided by experts in the field. Machine learning systems are achieving remarkable performances at the cost of increased complexity. Hence, they have become less interpretable, which may cause distrust.Asthesesystemsarepervasivelybeingintroducedtocriticaldomains,suchas medicalimagecomputingandcomputer-assistedintervention,itbecomesimperativeto develop methodologies to explain their predictions. Such methodologies would help physicians to decide whether they should follow/trust a prediction. Additionally, it could facilitate the deployment of such systems, from a legal perspective. Ultimately, interpretabilityiscloselyrelatedtoAIsafetyinhealthcare.Besidesincreasingtrustand acceptance by physicians, interpretability of machine learning systems can be helpful for other purposes, such as during method development, for revealing biases in the training data, or studying and identifying the most relevant data (e.g., specific MRI sequences in multi-sequence acquisitions). TheiMIMICproceedingsincludesix8-pagepaperscarefullyselectedfromalarger pool of submitted manuscripts, following a rigorous single-blinded peer-review pro- cess.Eachpaperwasreviewedbyatleasttwoexpertreviewers.Alltheacceptedpapers werepresentedasoralpresentations duringtheworkshop, with time for questions and discussion. We thank all the authors for their participation and the Program Committee mem- bersforcontributingtothisworkshop.WearealsoverygratefultooursponsorsH2O. ai, the Competence Center for Medical Technology (CCMT), Olea Medical, and the support of the promotion fund for young researchers at the University of Bern. September 2018 Mauricio Reyes Carlos A. Silva Sérgio Pereira Raphael Meier 1 https://imimic.bibbucket.io/.
The list of books you might like
