Improving CT Imaging with Artificial Intelligence

What biomedical issue did you address in your research, and what did your studies find?

Patients who are at risk for coronary artery disease often undergo a test known as a computed tomography (CT) scan to determine the presence of dangerous plaques that limit blood flow through an artery. However, conventional CT scanners have difficulty clearly seeing small or dense plaques and blockages and can cause plaques and stents to appear larger than their physical size.

Mayo Clinic was the first hospital in the world to install a new type of CT scanner, called a photon-counting-detector CT scanner, with better spatial resolution and the potential to visualize dense plaques, blockages and stents more clearly. It was approved for clinical use in 2021. My Ph.D. research focused on measuring the improvement of the new scanner in its ability to image previously hard-to-see materials. My first tests focused on objects of known size. Then, I moved on to compare the scans of clinical patients who received a coronary CT with both the conventional and the new scanner.

Our study found the new scanner provided more accurate results and was preferred by radiologists who interpret CT images. Because nearly all CT scanners still use the conventional technology, I developed a convolutional neural network (a type of artificial intelligence) to learn the resolution of the new scanner and apply it to the older systems. Our team calls the technique "Improved LUMEN visualization through Artificial super-resoluTion imagEs (ILUMENATE)." In a clinical evaluation of ILUMENATE, looking at 22 patients scanned on conventional CT scanners, radiologists consistently rated the ILUMENATE images superior to the original, conventional CT scans. Hence, the improved resolution and accuracy of ILUMENATE can provide better diagnoses in patients with dense plaques and stents.

What aspects of Mayo's culture helped you grow as a scientist and as a thinker?

Training in Mayo Clinic's collaborative research environment, I have appreciated the profound impact that working on a large, diverse interdisciplinary team has on patient care. My research group in the CT Clinical Innovation Center is comprised of a team of scientists, clinical physicists, CT technologists, and physicians. I had access to clinical conventional and photon-counting-detector CT systems, test objects to mimic patients for experiments, and micro-CT imaging. In addition to our lab's computing resources, I had access to larger, clustered computing resources for cloud-based deep learning environments.

In addition to my biomedical engineering and physiology coursework, I applied for and was fortunate to take courses at Mayo Clinic approved by the Commission on Accreditation of Medical Physics Education Programs (CAMPEP) to earn my certificate to apply for medical physics residency upon graduation.