Cloud computing has accelerated the physics simulations for RefleXion Medical, a medical device company that is developing a biologically guided radiation-therapy system for treating solid tumors.
RefleXion deployed CliQr Technologies‘ CloudCenter platform to shorten the time of physics simulations for tumor therapy treatments to overnight from two to three weeks, Akshay Nanduri, co-founder and vice president of business development for RefleXion, told eWEEK.
By running physics simulations in the cloud, RefleXion hopes to reduce the time to market for the device. With cloud computing, the company is able to pull data outcomes from scans in a timely manner, said Nanduri. The device, which requires approval by the U.S. Food and Drug Administration, will use high-dose radiation to try to eradicate cancer and will be used by academic medical centers.
“The results we got by using CliQr will be crucial as we start building this device,” said Nanduri. “It’s a tool we’re going to use throughout the development of this technology when we start building hardware.”
RefleXion’s technology uses Positron Emission Tomography (PET) during radiotherapy to enable non-invasive tracking and direct treatment of tumors.
PET scanning is a type of nuclear medical imaging that produces a 3-D image of processes in the body. “We’ve come up with a unique way to combine that with radiation therapy,” he said. “Radiation therapy is the way that we kill cancer, and PET imaging is the way that we detect cancer and find out where it is.”
RefleXion uses CloudCenter to run monte-carlo simulations, which are computational algorithms that rely on random sampling to obtain numerical results. The simulations are able to “mimic” how the PET-guided system will operate, said Nanduri.
After loading its simulation application, RefleXion was able to benchmark the software across several clouds to determine which platform would be the best fit.
“RefleXion Medical onboarded in a day and has since periodically benchmarked their application across different clouds in order to maximize their price-performance,” Gaurav Manglik, CEO and co-founder of CliQr Technologies, said in a statement.
After loading its application in CloudCenter, RefleXion completed a patient study using cloud-based software simulations. The study involved examining PET scans of lung cancer patients at Stanford Cancer Institute to see how the company’s proposed PET-guided device would perform on the data set of cancer patients.
“Initially we started out just with buying a couple of PCs and installing them in our office and running these simulations, which would take many days to run because they are pretty complex, scientific, physics-based simulations,” said Nanduri.
With software modeling and gathering of experimental imaging data, “we can have a very good sense of how our device is going to perform when it’s treating patients,” he said.
Without cloud computing, RefleXion would have have to purchase 20 to 30 servers to run locally. Pressure to get results out quickly to the scientific community would have led the company to an expensive investment in servers, said Nanduri.
RefleXion was able to run 30 Amazon nodes in parallel on the CliQr platform following some tweaks to the company’s software. “We wouldn’t be able to do that in the office—we don’t have 30 servers to run,” he said.
CloudCenter’s dashboard allows RefleXion to manage benchmarking, deploying and runtime of multiple applications on one or more clouds, CliQr reported.
In addition, the Orchestrator technology built in to CloudCenter enables this flexibility of being able to work with different clouds without recoding applications or developing cloud-specific scripts or virtual machine images, according to CliQr.
“Once you onboard your application, it’s like a single destination, a Web front end where in a menu drop-down I can choose a specific Amazon configuration to run my simulation,” Nanduri explained. “I can do the same with Rackspace or Google.”
Because Amazon, Google and Rackspace all have their own scripting languages, running RefleXion’s application on multiple cloud platforms would be too time-consuming, said Nanduri.