Beyond MIC is a a full-day workshop on integrating medical imaging and non-imaging modalities to answer novel clinical and healthcare challenges. External modalities comprise of a wide variety of heterogeneous sources of information with complex relationships and very large dimensionality. As current methods do not capture these complex interactions, joint representations and models are becoming important, making this an ideal workshop for presenting novel frameworks. We invite papers discussing novel methods with significant imaging and non-imaging components, addressing practical applications and new datasets. The following themes are emphasized:
The following is a tentative program and may change.
BeyondMIC will happen in the main conference center, in the room Picasso, at level -2.
Beyond MIC: a full-day workshop on integrating medical imaging and non-imaging modalities to answer novel clinical and healthcare challenges. Recent collaborative studies - such as ADNI, TCIA and the UK Biobank - and the hospital open-data initiatives are resulting in larger collections of medical images. Increasingly, these studies often also include non-imaging modalities such as clinical variables, electronic health records, insurance data, pathology reports, genomic data, and patient history. Alongside medical images, these rich external modalities present an opportunity for improving traditional medical image computing tasks like diagnosis, prediction, statistical analysis, identification or segmentation, as well as facilitate newer tasks like automatic image annotation. However, this extremely heterogeneous external data also poses unprecedented technical obstacles, including pre-processing different or inconsistent formats, modeling the complex noise and heterogeneity, jointly handling high dimensionality and complex structure in each modality, managing significantly larger amounts of big data, and developing efficient learning algorithms.
Machine learning methods tackling data-driven health care problems have been gaining interest, and workshops such as "Machine Learning in Health Care" and "NIPS Workshop in Machine Learning for Health" have boasted more than 300 attendees from various fields. MICCAI offers an ideal and timely opportunity to combine this rising interest in data-driven health care with medical imaging expertise. We will assemble researchers of different specializations and shared interests in this newly evolving field, facilitating the advancement of novel methods and technologies. Specifically, the mathematical, statistical and algorithmic thinking, and medical image analysis experience of the MICCAI community can help develop new methods for the analysis of emerging imaging and non-imaging modalities. Our workshop would therefore offer an ideal meeting to bridge the gap between the various communities that can contribute to these problems. Beyond MIC will include keynote sessions introducing the state of the art and challenges of the field, as well as presentations of accepted abstracts discussing novel methods or new applications.