George Sgouros is director of the Radiological Physics Division, where engineering, physics, and computational science come together to drive advances in imaging and therapy. Since joining the faculty in 2003 and stepping into the chief role in 2019, he has guided the division to expand and evolve, keeping pace with rapid changes in the field and building on its strengths.

While historically grounded in CT and cardiac imaging, the division now plays a central role in developing tools for precise therapy planning, including AI-enhanced image analysis, dual-photon CT imaging, and SPECT quantification. These innovations are tightly connected to clinical needs, supporting more precise therapy planning in areas like stroke evaluation, radionuclide therapy, and spectral CT.

Sgouros describes the division’s work as “bringing the tools of physics and computation directly to the bedside,” a philosophy driving both research and clinical translation. The division thrives on faculty exploring ways to automate, accelerate, and refine imaging for better patient outcomes, positioning Johns Hopkins as a leader in medical physics.

Transforming CT Imaging with Stationary Spectral Encoding

Pushing the boundaries of computed tomography (CT), Jingyan Xu, associate professor of radiology, is developing the stationary spatial spectral encoder (S3E), a device designed to enhance diagnostic accuracy and image quality in spectral CT. Now in the prototyping phase, the compact encoder can integrate into both energy-integrating and photon-counting CT systems.

With its stationary design, S3E handles spectral data acquisition while the traditional gantry manages tomographic imaging. This division of tasks could sharpen clarity and lower radiation exposure. “Our goal is to make spectral CT not only more precise but also more practical for a wider range of clinical settings,” Xu says. She collaborates with Yong Du, associate professor, and partners at the University of Utah to merge physics and engineering, driving innovation in CT technology.

AI-Powered Image Alignment in Cancer Therapy Monitoring

Junyu Chen, instructor, is developing advanced AI methods to improve image registration in nuclear medicine, helping track the effectiveness of cancer therapies. His work aligns sequential PET/CT or SPECT scans taken at different treatment stages to measure tumor response with greater accuracy. “By aligning these images, we can better see how a tumor is shrinking or responding to treatment,” Chen explains.

His AI models automate and enhance this process, providing more reliable comparisons without altering standard clinical workflows. While still in the early stages, this post-processing approach could reduce variability, increase speed, and support more precise decision-making. Chen’s efforts earned him a spot on Forbes’ 30 Under 30 list for Healthcare in 2024.

Transforming Stroke Care with 3D Imaging in the IR Suite

Focused on transforming stroke care, Ken Taguchi, professor, is bringing high-resolution 3D perfusion imaging into the interventional radiology (IR) suite. His imaging software reconstructs cross-sectional 3D views of tissue viability from standard procedures, giving radiologists clearer insight during critical interventions like clot retrieval.

Currently, IR teams rely on 2D X-ray shadows, which can limit visibility. Taguchi’s method aligns detailed pre-procedure scans with live imaging for real-time precision. “Right now, we’re using the system offline, but our aim is to move it into live clinical workflows so it can guide procedures as they happen,” he says. Collaborating with Vivek Yedavalli, chief of neuroradiology at Johns Hopkins Bayview, who provides clinical expertise, and Andreia Faria, associate professor of radiology, whose neuro–arterial mapping supports integration with complex anatomy, their work is redefining image-guided intervention to improve outcomes for stroke patients.