According to Bartron's president and CEO, Fitz Walker Jr., the company's Med- Seg application was developed using source code from a NASA earth science imaging program. The product has yet to be approved by the Food and Drug Administration (FDA) for use in medical practice. Walker said the company is currently in the final stages of submitting Med-Seg for that approval.
Once approved, Med-Seg will draw images from medical centers using a server placed at each medical facility. The information fed into that server will then be processed using the imaging program at Bartron's data centers and sent back out again to the hospitals for diagnosis. Medical centers will have the option to store the data long term at Bartron's facilities as well, Walker said.
Between those two events, the Kazeon software will have to correctly sort images according to content, feed each file into a high-performance computing (HPC) cluster for processing, detect when the HPC cluster sends the processed file to storage and then recall the file for sending across the WAN.
Bartron had searched for a product to handle those duties over the last several years, testing several options, including database management systems, and dismissing them all. Walker said the company was beginning development on a homegrown application when the company's storage vendor, IBM, which sold Bartron a rebranded Network Appliance Inc. (NetApp) FAS3020 filer, recommended Kazeon. (Editor's note: NetApp and Kazeon have a long-standing partnership; Walker said he dealt with Kazeon separately.)
"This was the first product I'd found that seemed like it could handle the workload," Walker said. He also said the product fit his needs because, like the Med-Seg application, it runs on Linux. Walker also liked the fact that the data classification appliance runs on commodity PC hardware and uses a Web browser for management.
Walker said Kazeon has also been willing to work with Bartron to "train" its data classification software to work with the Med-Seg application's algorithms, a tricky proposition since the licensing agreement with NASA stipulates that Bartron can't divulge source code to a third party.
Over the last month, the IS 1200 passed the first round of interoperability tests, which have been dry runs of the system's file layout on a small scale. This round of testing has also involved demonstrations of the data classification process and data migration at a jet propulsion lab to simulate the high-resolution images Med-Seg will produce.
The next round of testing will be formally documented for the FDA as the basis for the approval process, working with live data from hospitals, including the University of Connecticut Health Center, New York University Medical Center, Yale-New Haven Medical Center and the University of Maryland Medical Center.
Ultimately, each of the estimated 10,000 images Bartron predicts it will stream in each week from medical centers could be as much as 17 MB, Walker said. "And there can't be any delay or corruption -- it's safe to say we'll be putting it through its paces."
Walker said he's hoping that Kazeon will also be able to perform the conversion of files from the Digital Imaging and Communications in Medicine (DICOM) standard, a specialized transmission and compression algorithm for the transmission of medical images, to JPEG and TIFF. "We're working with another contractor right now to design an interface to convert DICOM," he said. "It would be nice to have the Kazeon box able to do that directly."
"Kazeon is adding TIFF conversion capability in the coming quarter, which will also be important for e-discovery," according to Michael Marchi, vice president of solutions marketing for Kazeon, in an email to SearchStorage.com. Kazeon has no plans to add JPEG conversion in the near future, he said.
If the Med-Seg system passes this round of testing, the next step is FDA approval and commercialization, a proposition that will multiply the workload for the system exponentially and require the construction of new data centers for Bartron. Once that happens, Walker said, Kazeon products will also be used to make and migrate copies of the data for disaster recovery purposes. A longer term project will be adding support for three-dimensional images, further increasing the workload.
"What we're envisioning is a kind of 'tele-medicine,' " Walker said. "We're especially interested in its application for the early detection of cancer, with fewer 'false positives.' "
Kazeon's competitors in the data classification space include Abrevity Inc., Arkivio Inc., EMC Corp., Hewlett-Packard Co. (HP), Index Engines, Intradyn Inc., Scentric Inc. and StoredIQ Corp. For more on data classification tools and vendors, please see our Buyer's Guide.