Using a science gateway to deliver SimVascular software as a service for classroom instruction

Abstract

SimVascular (http://www.simvascular.org) is open source software enabling users to construct image-based, patient-specific anatomic models and perform realistic blood flow simulation useful in disease research, medical device design, and surgical planning. The software consists of two core executables: a front-end application and a flow solver. The front-end application enables users to create patient-specific anatomic models from imaging data, generate finite-element meshes, prescribe boundary conditions, and set up an analysis. The finite-element based blood flow solver utilizes MPI and is massively scalable. SimVascular has been successfully integrated into graduate level courses on cardiovascular modeling at multiple institutions including Stanford, UC Berkeley, Purdue, and Marquette to introduce state-of-the-art modeling to the students and provide a basis for hands-on projects. While the front-end application can be installed and run on a laptop, the flow solver requires high performance computing (HPC) for realistic problem sizes. This provides a significant challenge for instructors as many students are unfamiliar with HPC, and local resources might be limited or difficult to administer. There is also a need to provide user and group management capabilities for courses: students should authenticate using campus credentials, instructors should be able to access students' work, and students' access to computing allocations should be limited. Our poster will detail an Apache Airavata-based science gateway to address these needs. XSEDE's Comet provides the backend computing power. This approach allows the SimVascular team to provision HPC resources and install and maintain the software providing students access at institutions across the country. The science gateway interface provides access to SimVascular's flow solver, while allowing students to use SimVascular's desktop interfaces.