Satellite Events

Published proceedings

LNCS 10541 - MLMI

All submissions to workshops and challenges will be managed independently by the respective organizers. The detailed information for each event will be announced on the event's website which is maintained by the event organizers.  A list of links to the individual websites will be maintained on this web page. Please note that the deadlines for events may differ. All participants are kindly requested to follow the individual events' web sites and contact the respective event organizers regarding event specific questions. Email addresses are found in the table below. For all other inquiries, please contact the Satellite Events Committee at satellites@miccai2017.orgworkshops@miccai2017.org,challenges@miccai2017.orgtutorials@miccai2017.org.

Workshops

Challenges

Tutorials

Satellite events descriptions

(W: Workshop, C: Challenge, T: Tutorial)

In medical imaging, augmented reality environments aim to provide the physician with enhanced perception of the patient either by fusing various imaging modalities or by presenting image-derived information overlaid on the physician’s view, establishing a direct relation between the image and the patient. This workshop is intended to bring together researchers in computer science, electrical engineering, physics, and clinical medicine engaged in the development of augmented environments for image-guided interventions.

Connectomics is the study of whole brain maps of connectivity, commonly referred to as the brain connectome, that focuses on quantifying, visualizing, and understanding brain network organization including their applications in neuroimaging. The primary objective of this workshop is to bring together computational researchers (computer scientist, data scientist, computation neuroscientist) to discuss new advancements in network construction, analysis, and visualization techniques in connectomics and their use in clinical diagnosis and group comparison studies. The secondary objective of this workshop to attract neuroscientist and clinicians to show recent advancements in connectomics, and how connectomics is successfully applied in various neuroimaging applications.

The goal of this workshop is to bring together imaging researchers and clinicians working in machine learning on multimodal data sets for clinical decision support and treatment planning to present and discuss latest developments in the field. Specifically, researchers interested in multimodal learning, biomedical imaging, medical image retrieval, data mining, text retrieval, and machine learning/AI communities will be co-located with clinicians who use computer-aided diagnosis and clinical decision support tools to not only discuss new techniques of multimodal learning but also their translation to clinical decision support in practice. We are looking for original, high-quality submissions that address innovative research and development in the learning of multimodal medical data for use in clinical decision support and treatment planning.

The main goal of BIVPCS is to continue the platform started in MICCAI 2013 (see: https://sites.google.com/site/mwbivpcs/) for communications among specialists from complementary fields such as signal and image processing, biomechanics, computational vision, mathematics, physics, informatics, computer graphics, bio-medical-practice, psychology and industry. The participants will present and discuss techniques and methods and explore the translational potentials of the related emerging technological fields. Therefore, BIVPCS is an excellent opportunity to refine ideas for future work and to establish constructive cooperation for new and improved solutions of imaging and visualization techniques and modelling methods toward more realistic and efficient computer simulations based on biomechanics principles.

The Brain Lesions (Brainles) workshop is a satellite event of the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI).
We aim at giving an overview of new advances of medical image analysis in glioma, multiple sclerosis, stroke and trauma brain injuries. We let researchers from the medical image analysis domain with radiologists and neurologists meet, and discuss these diseases and traumas, with the goal of comparing neuroimaging biomarkers which can be used on one or applied to the others.
The event will be held in conjunction to the challenges on Brain segmentation (BRATS) and White Matter Hyperintensity (WMH).

CARE 2017 will bring together researchers, clinicians and medical companies to advance the field of computer assisted and robotic endoscopy. This workshop will feature high quality, original papers and invited keynote presentations on the latest advances in developing the next generation of CARE systems.

Mathematical modelling and computer simulation have had a profound impact on science and proved tremendously successful in engineering. One of the greatest challenges for mechanists is to extend the success of computational mechanics beyond traditional engineering, in particular to medicine and biomedical sciences. The Computational Biomechanics for Medicine workshops provide an opportunity for researchers to present and exchange ideas on applying their techniques to computer-integrated medicine, including medical image computing, computer-aided modeling and evaluation of surgical procedures, and imaging, analysis methods for image guided therapies, computational physiology, and medical robotics.

Over the last two decades interest in diffusion MRI has exploded. The technique provides unique measurements sensitive to the microstructure of living tissue and enables in-vivo connectivity mapping of the brain. Computational techniques are key to the continued success and development of diffusion MRI and to its widespread transfer into the clinic. New processing methods are essential for addressing challenges at each stage of the pipeline: acquisition, reconstruction, modelling and model fitting, fibre tracking, connectivity mapping, visualisation, group studies and inference. This full-day workshop, now in the tenth edition, will give a snapshot of the current state of the art

CLIP’s focus is the effective translation of computational image-based techniques into the clinic filling the gaps between medical imaging, basic science and clinical applications. Submitted paper should be centered on specific clinical applications, including techniques and procedures based on comprehensive clinical image data. Submissions related to applications already in use and evaluated by clinical users are particularly encouraged. We welcome novel techniques/applications that use other image information (e.g. photographs, microscopy images or OCT) in addition to radiological image data. The integration of other non-image information (‘omics’ data, way-of-life information etc.) with medical image data is another focus of the workshop

 Molecular imaging is an evolving clinical and research discipline enabling the visualization, characterization and quantification of biologic processes taking place at the cellular and subcellular levels within intact living subjects. As a dedicated workshop, Computational Methods for Molecular Imaging (CMMI 2017) will cover various areas from image synthesis to data analysis and from clinical diagnosis to therapy individualization, using molecular imaging modalities PET, SPECT, PET/CT, SPECT/CT, and PET/MR. Technical topics will cover but not limit to image reconstruction, image enhancement, physiological modeling, computational simulation, multi-modal analysis, and artificial intelligence methods with clear clinical application and close industrial connection.

The two MICCAI-Workshop series (CVII and STENT) on technological and scientific research concerned with endovascular procedures are again brought together into a single workshop at MICCAI 2017. A continuous communication between surgeons/physicians and scientific and industrial researchers is crucial for bringing forward innovative solutions to the clinics: we aim to provide an interchange platform for medical experts and technological researchers concerned with all aspects of endovascular interventions. The workshop will focus on imaging and computed assisted technological advances that are poised to have dramatic impact on the diagnosis, analysis, modeling, and treatment of vascular diseases.

After the success of the 1st DLMIA and the 2nd DLMIA, held with MICCAI 2015 and MICCAI 2016, respectively, where we welcomed hundreds of attendees and influential invited speakers, we present the 3rd DLMIA to be held with MICCAI 2017. Deep Learning in Medical Image Analysis (DLMIA) is a workshop dedicated to the presentation of works focused on the design and use of deep learning methods in medical image analysis applications. We believe that this workshop is setting the trends and identifying the challenges of the use of deep learning methods in medical image analysis. 

This workshop will bring together the community of researchers working on younger cohorts and provide a forum for the open discussion of advanced image analysis of growth and development in the perinatal, preterm and paediatric period.

Graph-based models have been developed for a wide variety of problems in computer vision and biomedical imaging (e.g segmentation, registration, classification, shape modelling, population analysis). Different graph methodological aspects have been extensively explored, like graphical models, graph-theoretical algorithms, spectral graph analysis, graph dimensionality reduction, and graph-based network analysis. New topics are also emerging as the outcome of interdisciplinary studies, shedding light on areas like deep structured models and signal processing on graphs.
With this workshop we aim to bring together scientists working on both, well-established and emerging graph-based methods, and their applications to biomedical imaging.

The SWITCH workshop focuses on imaging related to stroke diagnosis and treatment. The main goals of the workshop are 1) to introduce the clinical background of challenges/opportunities related to imaging for stroke that are relevant for researchers working in the “MICCAI” field, and 2) to stimulate discussion and ideas exchange. To this end, there will be keynotes by clinical experts in stroke imaging and treatment, as well as presentations by researchers of on-going work. The tutorial character of the keynotes makes this workshop a perfect introduction or the ISLES workshop in the afternoon.

After the successful first LABELS workshop held at MICCAI 2016, we are happy to announce the second workshop on Large-scale Annotation of Biomedical Data and Expert Label Synthesis. Because machine learning approaches need large quantities of labeled data, the workshop focuses on acquisition of training data, design of labeling procedures and improving the annotation process for experts. Topics include active learning, semi-supervised learning, multiple instance learning, domain adaptation, transfer learning, crowdsourcing, fusion of labels, modeling label uncertainty, visualization and human-computer interaction.

MFCA is a MICCAI workshop devoted to statistical and geometrical methods for modeling the variability of biological shapes. The goal is to foster the interactions between the mathematical community around shapes and the MICCAI community around computational anatomy applications. The workshop aims at fostering interactions between researchers investigating the combination of geometry and statistics in the context of computational anatomy from different points of view. A special emphasis will be put on theoretical developments, applications and results being welcomed as illustrations. 

Imaging genetics studies the relationships between genetic variation and measurements from anatomical or functional imaging data, often in the context of a disorder. As large imaging genetics datasets are becoming available, their analysis poses unprecedented methodological and clinical challenges. The third edition of MICGen: MICCAI Workshop on Imaging Genetics (http://micgen.mit.edu) will bring together researchers and clinicians from various fields including medical genetics, computational biology and medical imaging, presenting a forum for both fundamental concepts as well as state-of-the-art methods and applications in this rapidly evolving field.

Machine Learning in Medical Imaging (MLMI 2017) is the eighth in a series of workshops in conjunction with MICCAI. This workshop focuses on major trends and challenges in the medical imaging field. It aims to help advance the scientific research within the broad field of machine learning in medical imaging. Topics of interests include but are not limited to (deep) machine learning methods with their applications to medical image analysis, registration, segmentation, classification, computer-assisted detection/diagnosis, multi-modality fusion, image reconstruction, image retrieval, cellular image analysis, and molecular/pathologic image analysis

Formerly known as Computational Spine Imaging (CSI) workshop, this year the scope is extended to computational methods and clinical application in musculoskeletal imaging. Giving the increasing volume of multimodal imaging examinations associated to musculoskeletal diseases and the complexity of their assessment, there is pressing need for advanced computerized methods that support the physician in diagnosis, therapy planning, and interventional guidance. In this workshop, we are inviting researchers to share and exchange their experiences and expertise in musculoskeletal computational imaging and modelling methods and applications. 

Age-related macular degeneration, diabetic retinopathy and glaucoma are main causes of blindness. Oftentimes blindness can be avoided by early intervention, making computer-assisted early diagnosis of retinal diseases a research priority. Related research is exploring retinal biomarkers for systemic conditions like dementia, cardiovascular disease, complications of diabetes. Significant challenges remain, including reliability and validation, effective multimodal analysis (e.g., fundus, optical coherence tomography, scanning laser ophthalmoscopy), more powerful imaging technologies, and the effective deployment of cutting-edge computer vision and machine learning. OMIA-4 will address all these aspects and more, this year in collaboration with the ReTOUCH retinal image challenge.   

We present the 3rd edition of the Patch-based techniques for medical imaging (PatchMI) workshop. This workshop is dedicated to the advance of scientific research within the patch-based processing of medical images. We welcome contributions related to patch-based processing in a variety of applications including (and not limited to): segmentation (e.g., brain, cardiac, MS lesions, tumors); enhancement (e.g., de-noising, super-resolution); computer-aided diagnosis; mono & multimodal registration; multi-modality fusion; mono & multi modal image synthesis; image retrieval; dynamic, functional & anatomic imaging; super-pixel/voxel-based analysis; sparse dictionary learning & sparse coding; analysis of 2D, 2D+t, 3D, 3D+t and 4D and 4D+t data.

We present the 3rd edition of the Patch-based techniques for medical imaging (PatchMI) workshop. This workshop is dedicated to the advance of scientific research within the patch-based processing of medical images. We welcome contributions related to patch-based processing in a variety of applications including (and not limited to): segmentation (e.g., brain, cardiac, MS lesions, tumors); enhancement (e.g., de-noising, super-resolution); computer-aided diagnosis; mono & multimodal registration; multi-modality fusion; mono & multi modal image synthesis; image retrieval; dynamic, functional & anatomic imaging; super-pixel/voxel-based analysis; sparse dictionary learning & sparse coding; analysis of 2D, 2D+t, 3D, 3D+t and 4D and 4D+t data.

The medical image community has always been fascinated by the opportunity to create simulated synthetic data upon which to develop, test, and validate image analysis and reconstruction algorithms. Mechanistic models simulating complex spatio-temporal acquisition processes and other dynamics have been a staple. Recently, machine learning-driven, phenomelogical models that can learn directly data associations have become popular. This workshop aims to invigorate research and stimulate new ideas on how to best proceed and bring these two worlds together.  We build on the success of SASHIMI 2016 where we had exceptional talks, a keynote speaker that excited the audience, and lively discussion.

The Challenge provides 120 cardiac CT/MR images, acquired in clinical environment, for various research groups to test and validate their methods, particularly for whole heart segmentation (WHS). The aim is not only to benchmark various WHS algorithms, but also to cover the topic of general cardiac image segmentation/ registration and modeling. The selected manuscripts will be published with the STACOM proceedings in Lecture Notes in Computer Science, Springer. In addition, the benchmarked algorithms will be summarized to form a full paper for potential publications in MedIA and IEEE TMI, and the participants will share the credit.

Point-of-care Ultrasound (POCUS) encompasses automated US image and RF data analysis algorithms, rugged US probes, robust tracking hardware, and specialized user interfaces including augmented reality systems.  The goal of a POCUS system is to guide novice users to properly manipulate an US probe and interpret the acquired data. The output of a POCUS system is typically a quantitative measure or an automated diagnosis, not a B-mode image.  POCUS applications range from detecting intra-abdominal bleeding at the scene of an accident to in-home monitoring of liver health. The POCUS workshop will feature invited and accepted presentations, live demonstrations, and a panel discussion. 

BraTS’17 utilizes multi-institutional pre-operative MRI scans and focuses on the segmentation of intrinsically heterogeneous (in appearance, shape, and histology) brain tumors, namely gliomas. To pinpoint the clinical relevance of this segmentation task, BraTS’17 also focuses on the prediction of patient overall survival, via integrative analyses of radiomic features and machine learning algorithms. BraTS’17 builds upon its 5 previous successful instances (BraTS’12-’16) and this year its datasets have been updated with more routine clinically-acquired 3T multimodal MRI scans and all its ground truth labels have been manually-revised by expert board-certified neuroradiologists.

The aim of the “Diffusion MRI Data Harmonisation” Challenge is the systematic evaluation of the performance of algorithms that enable the harmonisation of diffusion-weighted MRI data across different scanners. Harmonisation consists of making data sets acquired with different scanners/protocols as comparable as possible, and has become a pressing need in the era of Big Data. Combining data from several scanners would increase dramatically the statistical power and sensitivity of clinical studies, with obvious benefits in clinical trials and multi-centre research. This Challenge represents a first step to unlock this potential.

This challenge aims to directly compare WMH segmentation techniques. White matter hyperintensities (WMH) of presumed vascular origin are one of the main consequences of small vessel disease and well visible on brain MR images. Various automatic WMH segmentation methods have been developed, but it is hard to compare them. Each method is evaluated on a different ground truth (different number of subjects, experts, protocols) and using different evaluation criteria. This multi-institutional challenge provides a large dataset to evaluate the performance of various WMH segmentation methods.

CoronARe ranks state-of-the-art methods in symbolic and tomographic coronary artery reconstruction from interventional C-arm rotational angiography.  
Specifically, we will benchmark the performance of the methods using accurately pre-processed data, and study the effects of imperfect pre-processing conditions (segmentation and background subtraction errors). We will also study the effect of imperfect calibration on the reconstruction accuracy. The evaluation will be performed in a controlled environment using digital phantom images.

Computational Precision Medicine 2017 is a set of two distinct multi-institutional cancer imaging challenge competitions: (1) The liver radiomics challenge, where the goal is to establish new radiomic workflows and benchmark the predictive capacity of radiomics features, from correlation of liver metastasis computed tomography data with pathology and survival data; and (2) The digital pathology challenge, composed of two sub-challenges in classification and segmentation of nuclei in image tiles extracted from whole slide tissue images in four different cancer types.  The overall goal of the digital pathology challenge is to evaluate and compare performance of automated classification and segmentation algorithms. 

Endoscopic surgery is the preferred approach to many surgical procedures. As a result, endoscopic image processing and surgical vision are evolving as techniques needed to facilitate computer assisted interventions (CAI). However, what is missing so far are common datasets for consistent evaluation and benchmarking. As a CAI challenge at MICCAI, our aim is to provide a formal framework for evaluating the current state of the art, gather researchers in the field and provide high quality data with protocols for validating endoscopic vision algorithms. Similar to EndoVis 2015, the challenge consists of several sub-challenges based on a call for data.

This year the ISLES Challenge focuses on the prediction of tissue loss upon mechanical thrombectomy, as found on follow-up MRI imaging. Multi-spectral data along with clinical information will be made available, including manual lesion annotations drawn by experts. Research teams are invited to participate in the challenge and have the opportunity to evaluate their approaches on a common benchmark based on this high-quality clinical dataset.  ISLES 2017 aims to tie up with the success of previous years to bring together people interested in stroke imaging and medical image analysis.

The first year of life is the most dynamic phase of the postnatal human brain development. Accurate segmentation of infant brain MR images into white matter gray matter and cerebrospinal fluid in this critical period is of fundamental importance in studying the normal and abnormal early brain development. Many researches have been done on both neonatal and early adult-like brain MRI segmentation. To data, only few studies focus on the segmentation of 6-month infant brain images with the lowest tissue contrast. In this challenge researchers are invited to propose and evaluate their automatic algorithms on 6-month infant brain MRIs.

The Challenge provides 120 cardiac CT/MR images, acquired in clinical environment, for various research groups to test and validate their methods, particularly for whole heart segmentation (WHS). The aim is not only to benchmark various WHS algorithms, but also to cover the topic of general cardiac image segmentation/ registration and modeling. The selected manuscripts will be published with the STACOM proceedings in Lecture Notes in Computer Science, Springer. In addition, the benchmarked algorithms will be summarized to form a full paper for potential publications in MedIA and IEEE TMI, and the participants will share the credit.

The goal of AC-DC (Automatic Cardiac Delineation Challenge) is two-fold: (i) compare automatic segmentation methods of the left ventricular endocardium and epicardium and the right ventricular endocardium for both diastolic and systolic phase instances, and (ii) compare the performance of automatic methods for the classification of the examinations in five classes (normal case, heart failure with infarction, dilated cardiomyopathy, hypertrophic cardiomyopathy, abnormal right ventricle). This is done using one the largest public database of 3D clinical cine-MRI images acquired from 150 patients (30 per pathology plus 30 healthy subjects) all fully annotated by 2 clinical experts.

Given high resolution serial section Nissl stained images of whole brains, the challenge is to produce a segmentation of brain regions on each section. The goal of the challenge is to evaluate automatic neurohistological tissue segmentation methods with respect to an ontology of expert-annotated brain region labeling. The challenge is an activity of The Brain Architecture Project, striving towards achieving comprehensive neuroanatomical understanding of vertebrate brains.

Due to their heterogeneous and diffusive shape, automatic segmentation of tumor lesions is very challenging. With our challenge we encourage researcher to develop automatic segmentation algorithms to segment the liver and its lesions to finally assess the tumor load in contrast¬-enhanced abdomen CT scans. The data and segmentations are provided by several clinical sites, scanners and protocols. The challenge is organized by the image-¬based ¬biomedical modeling Group (IBBM) and the Institute of Radiology of Technical University of Munich, Ludwigs Maximilian University, Radboudumc, Polytechnique Montréal, Tel Aviv University and IRCAD.

Optical coherence tomography (OCT) is becoming the most important diagnostic modality for imaging the retina. The presence of retinal fluid in OCT is an important indicator of devastating eye diseases such as age-related macular degeneration or retinal vein occlusion. RETOUCH Challenge consists of two tasks: (1) detection and (2) segmentation of intraretinal, subretinal and pigment epithelial detachment fluid. The goal is to compare automated algorithms that are able to detect and segment the three types of fluid on a common dataset of OCT volumes representing different retinal diseases, acquired with devices from different manufacturers.

The motivation for this tutorial is to facilitate the translation of algorithms to clinical environments, so their benefit to patients can be fully explored. Skills required for clinical translation are significantly different from those required for algorithm development. Open-source platforms have been developed for making the process easier. In the first part of the tutorial, invited speakers will give an overview about open-source platforms and their vision for the future of translational clinical research. In the second part, the audience will build a working brain surgery navigation system using devices provided by the organizers.

The objective of this tutorial is to introduce the MICCAI community to the new kinds of DICOM objects that can be used for storage of the data typically produced in the process of quantitative image analysis. Over the course of presentations and hands-on sessions, we will explain how DICOM can, and perhaps should, be used for storing your processing results such as segmentations, parametric maps and volumetric measurements. After completing this tutorial, attendees are expected to develop an understanding of the relevant new capabilities of the DICOM standard, as well as the tools that they can use to experiment with adoption of the standard in their everyday research.

Proper benchmarking of image analysis algorithms makes life easier not only for future developers (to learn the strengths and weaknesses of existing methods) but also for users (who can select methods that best suit their particular needs). The tutorial will concentrate on best practices to design biomedical image analysis benchmarks and challenges to measure algorithm performance in a standardized way. First, the design of a benchmark or a challenge will be presented including proper selection of datasets, tasks and evaluation metrics. Next past benchmarks and challenges will be shortly reviewed. Finally, the topic will be summarized and future directions discussed.

Deep learning methods have recently become popular in medical imaging systems, such as for the segmentation of various types of tissues in medical imagery. In this tutorial, we will provide an introduction to deep learning, covering both theory and practice. On the theory side, we will describe the most common concepts found in today’s deep learning research, with a focus on convolutional neural networks. On the practical side, we will describe how the TensorFlow library can be used to apply deep learning in medical imaging, covering both TensorFlow basics and specific use cases in medical imaging applications.

Important dates

Copyright

Authors will assign to MICCAI Society the right to distribute the workshop / challenge / tutorial material to MICCAI members and to workshop / challenge / tutorial and conference attendees, independently of the copyright ownership planned by the workshop / challenge / tutorial organizers. It is up to the workshop / challenge / tutorial organizers to collect the agreement of the authors for that and get consent from the journals/publishers utilizing the aforementioned material from MICCAI workshops / challenges / tutorials.

Logistics and Budget

Organizers of Satellite events can expect the following logistical and financial support from the MICCAI 2017 organizing committee:

  • Satellite event room, of sufficient capacity (ranging from 25 to 250 participants);

  • Audio-visual support for each event: one Windows laptop, projector, audio, and technical support;

  • Refreshment breaks and lunch provided to all workshop participants;

  • Registration reports for each event at three time points before the event;

  • Three (3) registration fee waivers for the Satellite event. These do not apply to the main conference. Organizers for events will distribute these waivers at their discretion; and

  • Dissemination of the Satellite events PDF articles to participants, provided these are submitted on time and within guidelines

  • Free internet access for all participants

 

The MICCAI 2017 organizing committee will not support travel arrangements or stipends of the organizers of workshops / challenges / tutorials.

Please note that participants at MICCAI 2017 will be asked to register for Satellite events separately from the main conference, on a day by day basis.

Events organizers are responsible for addressing the following logistical challenges:

  • Submission and event website

  • Speaker presentation support

 

Sponsorship of the workshops should be coordinated with the General Chair of the MICCAI 2017 conference, in order to maximize the opportunities for the organizers, the companies, and all participants.