Large-Scale Computing Frameworks for Developing Image Analysis Tools

Steve Pieper
Harvard Medical School

The National Alliance for Medical Image Computing (NA-MIC) has been chartered to create an open source software infrastructure to support translational biomedical imaging research. Within NA-MIC and a number of closely collaborating groups, we have assembled software to address major building blocks including file I/O, numerics, visualization, user interface, and data archiving which are collectively called the NA-MIC Kit. For all these tools we have invested heavily in interoperability and feature enhancement while in some cases we have written new tools from scratch to fill important functional blocks. In addition, we promote a software engineering methodology around the themes of open development, ongoing cross-platform testing, and coding conventions that promote maintainability and code reusability. In addition to the core tools, we have created domain-specific workflows to address biomedical research areas from population studies of neurodegenerative diseases to robot guided prostate interventions. Addressing these clinical use cases helps refine and validate the underlying software systems. In this talk I will review the motivations, organization, and implementation of the NA-MIC Kit with particular emphasis on our efforts to build a distributed community of like-minded developers and users who share not just a common programming environment, but a common language for effective collaboration. I will review how these efforts are beginning to have a positive impact on particular research areas and how we plan to accommodate anticipated growth of the community.

Audio (MP3 File, Podcast Ready) Presentation (PowerPoint File)

Back to Long Programs