High-def Ultrasounds Improve Breast Cancer
and Other Imaging-Based Diagnoses

PDF of Case Study

The Challenge

Ultrasound scans have helped save tens of thousands of lives by catching early stage breast cancers. But the imaging technology is far from flawless, often leading to false alarms or worse, undetected tumors. Conventional beamforming algorithms have been used in ultrasound scanners for nearly a half century to help diagnose breast cancer and other life threatening conditions. The problem is that they typically result in degraded images that are blurry or cluttered. The culprit: off-axis signals, or the sound wave reflections coming from undesired locations within the organ or tissue.

Biomedical engineers at the University of Virginia (U.Va.) School of Engineering and Applied Science set out to develop a new algorithmic approach to solve the problem. Led by Associate Professor William F. Walker, the biomedical engineering research team created an advanced beamforming algorithm –called the Time-domain Optimized Near-field Estimator (TONE)–which significantly improves the contrast and resolution of ultrasound images.

The TONE algorithm reduces undesired off-axis signals, resulting in much higher definition images, but at the price of a much greater computational load. The algorithm developed on desktop computers overwhelmed the computer's processing ability. The team needed to parallelize the code and tap the University's high-performance computing (HPC) resources to make TONE practical for real-world diagnostic applications.

Star-P Solution

Using Star-P®, the biomedical engineering research team was able to automatically parallelize the TONE algorithms to run on a powerful, parallel system, without having to rewrite the code in C and MPI. They can now code algorithms and imaging models on their desktops using MATLAB®, but run them instantly and interactively on a 32-processor Linux cluster with 64 gigabytes of memory. The result is the ability to generate ultra-high-definition images that may someday revolutionize medical ultrasounds, potentially leading to more accurate and timely diagnoses of breast cancer...and more.

"We were not able to generate images with such a fine sampling pitch until we used Star-P," says Research Associate Francesco Viola. "It takes a huge amount of memory and computational resources to execute the algorithm. Typical resolution for ultrasound imaging systems is in the 200-300 micron range. With Star-P, we were able to tap into the University's supercomputing clusters to generate ultra high resolution images of 67 microns, without having to become parallel programming experts."

The TONE research project was funded by a grant from the U.S. Army Congressionally Directed Medical Research Program in Breast Cancer. But the algorithm's flexibility has potential to improve research in a host of other disciplines as well, according to James H. Aylor, dean of U.Va.'s School of Engineering and Applied Science. "The potential applications for this algorithm are almost infinite," he said. "Not only can it be used in the medical community to benefit patients nationwide, it will have applications in the fields of radio astronomy, seismology and more."

Summary & Metics

• Improved image resolution 4-fold: from hundreds of microns, to just 67µm
• Eliminate need to program MATLAB scripts in C and MPI
• Imaging breakthrough applicable to medical research, astronomy, and seismology