ACADEMIA

The Bartron Medical Imaging has selected Scali Manage to implement and manage a 64-node Hierarchical Image Segmentation Cluster, which is the first of ten Linux clusters that are planned for implementation in the United States. The clusters perform the processing of digital images, used by diagnosticians, via Bartron’s MED-SEG Systems that will aid in the early detection and treatment of potentially life threatening diseases. The initial cluster is expected to process a minimum of one terabyte of data every 15-20 days with an average job time of less under one minute. The cluster will be centrally located to receive, process and provide Medical Regional Hierarchical Segmented data. Bartron’s MED-SEG Systems will be installed at Hospitals, Radiology Centers and Universities in dispersed geographies. The MED-SEG system assists diagnosticians in the diagnosis of diseases that are imaged using Digital X-Rays, Soft Tissue, Mammograms, Ultrasounds, MRI images and CAT scans. Scali Manage will be used to monitor the performance of the cluster to ensure that it is operating at peak performance with maximum uptime. The cluster management software will also provide real-time metrics that indicate how the cluster is performing, allowing Bartron to make adjustments as needed. This will ensure that the cluster is able to handle the intense compute demands expected to flow from the MED-SEG products. “The data analysis demand on our cluster will be intense and that intensity is magnified when you think about the life altering importance for the diagnostician of getting the analysis right,” said Fitz Walker, President and CEO, Bartron Medical Imaging, LLC. “In addition, we are in the process of preparing our Device Submission for the Food and Drug Administration (FDA) review, clearance and authorization to market the product. Professionally developed and supported cluster management software was a must.” Scali’s hardware and software agnostic approach was also important for Bartron as the organization didn’t want to be locked into one vendor or technology, but rather, have the flexibility to take advantage of the latest technology to meet its growing performance requirements. “Organizations that deal with complex analysis coupled with enormous amounts of data are open to complications with their clusters,” said Bjørn Skare, president and CEO of Scali. “In Bartron’s case, not only does the cluster have to be working at peak capacity, it also has to be available 24 hours a day, seven days a week. Scali Manage eases the burden of managing the cluster and provides actionable intelligence that ensures the system is optimized for performance and availability.” “Bartron expects to have 10 additional clusters throughout the U.S. in the next three to four years. With these expansion plans, we needed a cluster management system we could standardize on and trust to continue to deliver industry leading functionality. Scali demonstrated to us that they had the products, expertise and support organization we needed going forward,” continued Walker. “In the field of medical analysis, the objective is zero margin for error so we need a technology partner like Scali that we can rely on.” In conjunction with Scali Manage, Bartron will also utilize PBS Professional from Altair for workload management. The out-of-the-box integration with Scali Manage is vital to ensure appropriate resources are allocated to each job, enhancing cluster performance and data compute speed. Astaro Security Gateway software will be implemented to maintain the security of the network. The cluster, which is scheduled to be installed and functional in July 2006, is made up of Intel-based servers using Gigabit Ethernet connection. This cluster was designed specifically for medical image processing utilizing Hierarchical Segmentation software licensed from NASA Goddard Space Flight Center. Additional information about Bartron Medical Imaging, LLC is available at its Web site Additional information about NASA Goddard's technology transfer program is available at its Web site.