Academic and industry leaders team up to develop next generation medical imaging technology

Conventional CT imaging. Filtered back projection (FBP) image reconstruction (left). Image reconstruction using Veo (right).

Traditionally in CT, physicians have had to balance the desires for high image quality and low radiation dose levels. While high image quality often required greater patient exposure to diagnostic radiation, lower dose levels for the patient usually meant lower image clarity from higher noise and more artifacts. Veo helps to change the equation, and takes the next step toward GE’s goal of routine CT imaging at under 1 millisievert (mSv) per exam with profound image clarity. In comparison, traditional CT exams, depending on the exam and patient, expose the patient to anywhere from 2-15 mSv of effective radiation dose, while natural background radiation exposes the average American to around 3 mSv of radiation per year.^†

Veo’s advanced algorithm was crafted by Charles A. Bouman, Ph.D., professor at Purdue University, and Ken Sauer, Ph.D., associate professor at Notre Dame, in collaboration with Jiang Hsieh, Jean-Baptiste Thibault and Zhou Yu from GE Healthcare.This collaborative research began in 2001, before iterative reconstruction was a major focus of medical imaging manufacturers. Ten years later, in late 2011, these three organizations signed an exclusive license agreement allowing Veo and the underlying MBiR technology to be commercialized by GE Healthcare.

“Conventional CT scanning takes thousands of views from different angles to 'see' 3D structures inside the body," Dr. Bouman says. "However, Veo can form the same 3D image with much less X-ray exposure, or dosage, because it uses a computer algorithm called Model-Based Iterative Reconstruction (MBiR). MBiR solves an enormous puzzle to find the best 3D reconstruction so it can both reduce noise and create a much sharper image.”

Scientists from the University of Michigan (U-M) also contributed to the research behind Veo, and the U-M Health System recently became one of the first U.S. teaching hospitals to install Veo – enabling some CT scans using a fraction of the dose required for a conventional CT image.

Jeff Fessler, Ph.D., a U-M professor in the Electrical Engineering, Computer Science, Radiology and Biomedical Engineering departments collaborated with GE on the project and explains how Veo is a new way of processing data. Using the MBiR technique, the technology employs sophisticated algorithms to produce clearer information from the existing X-ray data.  A CT scan typically consists of thousands of cross-section images of the body that must be reconstructed from raw data. “It’s a staggering amount of data,” says Fessler. “And doctors can’t just look at the raw data; they need software to process it into a meaningful image.”

Once Veo’s revolutionary algorithm was established, GE partnered with Intel engineers to speed up CT image processing time from several days for one clinical case to now as fast as multiple cases in one hour. Intel’s processor performance helped allow increased throughput and, by extension, staff productivity - and ultimately helped make Veo practical in a variety of clinical settings.

“Using Intel® Xeon® Processor technology in commercial medical solutions, GE Healthcare has been able to achieve diagnostic images while helping to lower patients’ radiation dose levels significantly.” says Michelle Tinsley, Intel Personal Solutions General Manager.

Veo’s development and clinical potential reflect how GE Healthcare continues to work alongside leading academic institutions, cutting edge industry partners and healthcare professionals globally to find new ways to drive better outcomes and improve patient experience.

* Trademark of General Electric Company.

^ In clinical practice, the use of ASiR and Veo may reduce CT patient dose depending on the clinical task, patient size, anatomical location and clinical practice. A consultation with a radiologist and a physician should be made to determine the appropriate dose to obtain diagnostic image quality for the particular clinical task.

^†Statistics from RadiologyInfo.org