Background and Scope

Modern learning algorithms make the promise of bridging the semantic gap between images and diagnoses and even reaching superhuman performance. The goal of this workshop is to explore the use of “big data” algorithms for harvesting, organizing and learning from large-scale medical imaging data sets and for general-purpose automatic understanding of medical images. This includes modern, scalable and efficient algorithms for automatic localization, segmentation, registration and characterization of anatomical features and anomalies. We encourage the submission of original papers that propose new methodology strongly motivated by a clinical application. Submissions will be at the interface of big data algorithms, computer vision, machine learning and medical image analysis. 

Possible topics

• Computer vision approaches that are scalable to big data
• Methods that are designed for efficient execution in a cloud environment
• Methods dealing with incomplete-, weak- or noisy annotation of training examples
• Data driven and exploratory models for image segmentation and quantitative description
• Learning approaches for registration, calibration and related image transforms
• Generative models of 3D image scenes relying on, or complementing, population atlases of anatomy or function
• Anatomical structure localization through object recognition and categorization
• Features and algorithms dealing with image acquisition variations, such as CT scan plan or MR pulse sequence variations, with/without contrast agents

• Semantic anatomy parsing, semantic navigation and visualization
• Applications of web-driven techniques to structure medical data sets
• Image indexing, data organization, and data harvesting
• Real-time medical image applications
• Algorithms using or evaluating big data sets, such as the VISCERAL data set