Pushing the Boundaries of Medical Imaging
May 2015—CT and MRI scanners are workhorses of modern medical imaging, yet their large size, expense and limitations in imaging dynamic structures like the heart limit their capabilities. By pushing the bounds of those technologies to make them more versatile, researchers at the Johns Hopkins University School of Medicine are changing how medicine is practiced.
About 15 years ago, Jeff Siewerdsen, a physicist and biomedical engineer, began adapting emerging technologies in digital X-ray imaging to CT. “The technology behind conventional CT is fantastic, and last year marked the 25th anniversary of breakthroughs that have made it a mainstay in diagnostic imaging,” he says. “But CT scanners may not be well-suited to applications requiring open access to the patient and a small—even mobile—footprint.”
For image-guided interventions, that is an important consideration. For example, one of Siewerdsen’s first projects was to create a CT system for guiding radiotherapy of tumors. He took advantage of new (at the time) detector technology that provides high-quality, large-area imaging in real time. The resulting technique is often referred to as cone beam CT, because the X-ray beam forms a volumetric cone rather than a narrow fan, as in conventional CT. Siewerdsen was on the team that incorporated cone beam CT into a medical linear accelerator, which generates a highly targeted beam of radiation to kill tumors. CT guidance helps to minimize sources of geometric error in targeting the tumor and is now the state of the art in image-guided radiation therapy.
Siewerdsen then began implementing cone beam CT on a mobile C-arm, a C-shaped medical imaging device already present in many operating rooms. His work helped to advance the capability of C-arms from taking simple, two-dimensional X-rays into three-dimensional cone-beam scanners that can provide a surgeon with a high-quality, real-time image of a patient during surgery.
Most recently, Siewerdsen worked with industry research collaborators to create a compact, mobile cone beam CT scanner that can be oriented to image a patient’s hand, wrist and elbow, but that can also be flipped to image a standing patient’s foot, ankle and knee. Such capability could provide new insight on injuries and conditions like arthritis that are better depicted when the body is bearing a natural load.
Researchers working with Siewerdsen also have been improving the software used to crunch CT data. Until recently, scanners have reconstructed images in a simple way that can be susceptible to image noise and artifact. “This poses basic limitations on image quality,” Siewerdsen says, “especially at low dose.” For instance, if a person has a small hemorrhage in the brain, image noise could make it difficult to detect. But a new process uses a model of the statistical properties of image noise to reduce noise while preserving resolution. Such methods are computationally intense, but modern high-speed computing hardware, based in part on graphics processing units arising from the video game industry, have made such tasks practical. The method enables high-quality CT images to be made using fewer X-ray projections, reducing exposure to radiation.
Those dual improvements to hardware and software are enabling Siewerdsen’s latest project: building a portable CT scanner to allow reliable, high-quality imaging and detection of brain injuries in the intensive care unit, urgent care, ambulances and even locker rooms. Currently, receiving a CT scan requires getting to a caregiver’s office, which can take a long time. That delay means that milder injuries—the most common type of head injury—may go undetected, which can result in a serious cascade of debilitating brain trauma. Siewerdsen’s goal is to enable patients to receive the most effective therapy at the earliest possible point following injury, rather than bear the burden of undetected disease and re-injury.
Imaging Moving Body Parts with MRI
Meanwhile, Daniel Herzka, an MRI physicist and biomedical engineer, has been working to improve MRI technology. While CT scans are great for imaging harder structures and some soft tissue, MRI scans are best suited for distinguishing among soft tissues in the body. However, says Herzka, “MRI scans are slow relative to the pace of our physiology.” That is, they perform poorly when imaging anything in motion, such as hearts, moving joints and unruly children who won’t lie still. As a result, doctors have had to compensate in less than ideal ways. For example, those requiring cardiac images are often asked to hold their breath for several seconds during a scan to stop breathing motion, a challenge for people with heart problems. “Our aim is to make imaging faster, easier and more informative for clinicians,” Herzka says.
MR images of a heart using today’s standard clinical technology (left) and high resolution techniques (right).
As with CT scans, MRI scan reconstruction and analysis has emerged as a hot area of research. Because MRI scans are especially good at generating contrast, cardiologists can acquire precise images by specifying two elements in the body to compare, such as the heart muscle and the blood in the heart. They can also change the machine’s settings to obtain images with different contrasts, and then select the image that provides the clearest picture. Moreover, thanks to new and better algorithms capable of sifting through noisy data, it’s now possible to approximate what parts of the image were skipped during a fast scan. Herzka is working on new fast and robust ways to get images that provide more accurate information about the patient by identifying the different types of tissue present in injured tissues of the heart.
Herzka, along with cardiologists at Johns Hopkins, is now using pigs to test the use of new MRI scans for patients undergoing electrophysiological intervention. Some patients, particularly those who have experienced a heart attack, develop scar tissue on the heart that can lead to irregular heartbeats, Herzka explains. To treat them, doctors typically use X-ray fluoroscopy in combination with electrical activity to guide them as they thread catheters through blood vessels and into the heart to ablate the areas of scarring suspected of causing problems. But X-rays, which are not ideal for soft tissues, produce shadows of the heart, making it hard to see exactly where the catheter is located. “Nothing can match MRI in its ability to discriminate between tissues types, such as scar, normal heart muscle and water accumulation due to injury,” says Herzka. It is not surprising that 50 percent of patients require further treatment, including more procedures, or develop new electrical problems as a result of treatment. “The doctors just can’t see well enough.”
In this experiment, pigs receive an MRI of their heart before the procedure. Then, the electrophysiologist standing just outside the machine uses real-time MRI scans to thread the catheter to the right location. Finally, the pigs undergo another MRI after the procedure to check if the ablation worked as planned. “By developing new ways of looking at tissue with MRI before, during and after intervention, I think we can dramatically improve upon the success rate of this and similar procedures in the near future,” Herzka says.