by Hector Marinez

I’ve been at NVIDIA for almost two years, and I often hear about the different industries our technology is changing. But when I found out how GPUs are being used to improve breast cancer screening and diagnosis, it really hit home.

A big reason for my interest is that my wife is at high risk for breast cancer, due to her family history, and has to deal with yearly screenings at a relatively young age.  Changes are coming that will make these events less anxious, not only for her but for the rest of our family, as well.

Breast cancer is the most common form of the disease among women in the U.S. and the second deadliest kind of cancer. Every year in the U.S., over 55,000 women die from breast cancer, and at least 10-20 percent of those who are screened do not detect existing cancer masses partly due to the presence of dense breast tissue.

Despite all this, there has been little recent innovation in breast cancer screening. Mammography, sonograms, MRI and self-examination are standard practices, but each is limited. They can entirely miss cancers, require lengthy examinations or can be painful for the patient.

Researchers at the UCSD Moores Cancer Center, in San Diego, are in clinical trials (150 patients have already been tested and another 130 are being screened at the Mayo Clinic in Rochester) of a promising new CUDA GPU-powered diagnostic machine that could radically improve the breast cancer screening process. It’s based on technology developed by Salt Lake City’s TechniScan Medical Systems, which allows patients to comfortably lie on their stomachs and have the entire breast – dense or not – scanned with minimal contact through automated whole-breast ultrasound. TechniScan provides a high-resolution, 3D-gamer quality image of the breast immediately – often within a 30-minute office visit – so patients spend less time waiting for test results.

By using CUDA to power the imaging process, technicians can produce 3D pictures that are capable of dividing the breast tissue into individual layers, making it harder for cancers to hide. Technicians are also able to produce such clear images that they can see cells thicken, an early process in the forming of cancerous cells, so they can monitor and diagnose a patient at an earlier stage of breast cancer. This is key to drastically increasing survival rates.

In addition to TechniScan, Massachusetts General Hospital is utilizing CUDA for digital breast tomosynthesis, which enables doctors to get a 3D visualization of the breast. And the University of Wisconsin-Madison is using CUDA for a new mammography technique that produces 3D images using low-power electromagnetic waves.

As I hear more about FDA approvals, I will provide you with updates. But for now, my wife and I are certainly rooting for these innovations to make it to all screening facilities soon.