The Edward S. Rogers Sr. Department of Electrical and Computer Engineering at the University of Toronto Uses CAPI-enabled Power Systems to Improve Cancer Treatments
Jeffrey Cassidy Ph.D. candidate with the University of Toronto ECE - Photo by Michelle Gibson
|Customer: The Edward S. Rogers Sr. Department of Electrical and Computer Engineering at the University of Toronto
Business: Various research projects, including cancer treatment
Challenge: Speeding up simulations for improved photodynamic therapy cancer treatment
Solution: Developing a field-programmable gate array (FPGA) running on an IBM Coherent Accelerator Processor Interface (CAPI) linked to a POWER8 processor
Hardware: An IBM Power Systems server running CAPI, as well as a custom-built FPGA chip
Software: Ubuntu Linux
IBM Power Systems* servers are mostly associated with business, running core production applications that support everyday operations. But because of their flexibility, speed and overall affordability, other segments—including medicine—are quickly adopting the technology.
The Edward S. Rogers Sr. Department of Electrical and Computer Engineering (ECE) at the University of Toronto, for example, uses Power Systems technology to collaborate with pharmaceutical companies and various medical centers to improve photodynamic therapy (PDT) for cancer treatment. In this case, it’s using a field programmable gate array (FPGA) chip attached to the IBM POWER8* Coherent Accelerator Processor Interface (CAPI) to speed up clinical PDT simulations.
“What got us onto the POWER8 platform was the CAPI link and the supporting intellectual property. It was really effective for us because there’s no other platform that has the same level of support for FPGA and processor integration.” —Jeffrey Cassidy, Ph.D. candidate, University of Toronto ECE
Formerly running on an x86 platform, the department’s simulations have since measurably accelerated at a considerably reduced cost. As Jeffrey Cassidy, Ph.D. candidate with the University of Toronto ECE, explains, “We have the potential to go up to 16 times faster per FPGA chip versus per x86 socket. And power efficiency is on the order of 40 times more compute per amount of power consumed.”
The Optimal Effect
Those numbers, however, belie the importance of the work by Cassidy and his collaborators. Rather than relying on the multi-angled approach of some forms of cancer treatment, including radiation, chemotherapy and even surgery, PDT is much more selective when targeting tumors.
To that end, it involves the use of nontoxic drugs that circulate throughout the body and only become active when exposed to light of an appropriate wavelength. When the drugs absorb a photon, they generate radicals that damage or, at the appropriate dose, kill cells in the targeted region.
PDT is currently in widespread use in parts of Europe and South America to treat skin cancer. These tumors tend to be close to the skin’s surface and typically not near vital organs. In this case, the photosensitizer is applied as a topical cream to the cancerous lesion and then exposed to the proper light wavelength from the surface.
Treating deeper-seated cancers, such as those of the lungs, head and neck, and bladder, however, require a different technique, one that’s highly targeted and avoids damaging surrounding tissue. And that’s where Cassidy’s research, in conjunction with organizations such as Princess Margaret Cancer Centre in Toronto, the Hospital Albert Einstein in Sao Paulo, Brazil, and the Roswell Park Cancer Institute in Buffalo, New York, comes into play.