Magnetic resonance imaging (MRI) is a non-invasive imaging technique widely used by clinicians to examine a patient’s organs, tissues, and skeletal system and diagnose conditions. At GW Engineering, Assistant Professor of Biomedical Engineering Junghun Cho and his lab students are advancing MRI-based techniques to better understand the human brain.
Cho and four Ph.D. students, Liukailai Ding, Tian Qiu, Renlong Yang, and Arpita Misra, joined professionals from around the world at the largest meeting dedicated to MRI, the International Society for Magnetic Resonance in Medicine (ISMRM) Annual Meeting, to share their work. Together, they presented 13 abstracts during this six-day event in Cape Town, South Africa.
Ding and Qiu, both in the second year of their GW doctoral program, were among only 131 individuals selected to receive Summa Cum Laude awards for their presentations. Their research focuses on improving the accuracy of methods to quantify cerebral oxygen extraction fraction (OEF)–the rate at which the brain utilizes oxygen–using artificial intelligence and machine learning.
“Receiving this award is a huge honor and a meaningful encouragement at this early stage of my PhD. I feel that all my effort has paid off. Professionally, this recognition from the ISMRM community makes me feel my work is valuable and encourages me to continue working in this field, which can provide clinically useful tools for brain physiology, aligning with my initial motivation for pursuing a PhD,” said Ding. “I am also very grateful to my advisor, Dr. Junghun Cho, for his guidance all the time and to other lab members for their assistance.”
The OEF is an important biomarker for several neurological diseases, including stroke, multiple sclerosis, and dementia. However, no current mapping technique is routinely used in clinical settings.
In his research, Ding proposed QQ-S, a deep learning method for estimating OEF using MRI. This improves upon the previous model the group developed, QQ-Net, which set a precedent for solving QQ-based biophysics models using deep learning. Results show that QQ-S better captures signal changes using a routine MRI sequence and demonstrates improved accuracy in estimating abnormal oxygen metabolism, especially in lesion regions.
Qiu’s research addresses a common challenge across deep learning-based OEF methods: ensuring that models can generalize across patients and imaging conditions for future clinical use. In his presentation, he introduced a new module, QQ-F, which eliminates the need for patient-specific fine-tuning by extracting high-dimensional, physics-based features from raw MRI signals that are then used as network inputs. Now, when inference is done for new subjects, they only need to extract those features rather than retrain the entire network.
“This is the first project I led during my PhD, so receiving this award is really meaningful to me,” said Qiu.
Qiu joined Cho with one other student to attend the meeting in person, while others attended virtually. Regardless of the format, each student found participating valuable as it offered them an opportunity to learn about the latest advancements in MRI research in their area of study and related fields.
“Some of the research was directly related to my work, some was not, but all of it helped me better understand the latest developments in the MRI field,” said Qiu. “It also inspired me to think about how new ideas and methods could be applied to my own research. It felt like brainstorming every day–challenging and tiring, but absolutely worthwhile.”
On the topic of neuroimaging of traumatic brain injury, white matter, and gray matter, Cho moderated a panel alongside Dr. Ricardo Coronado-Leija of New York University. The session featured abstracts spanning methodological developments and biological applications of state-of-the-art MRI techniques to explore microstructural, functional, and connectivity-related brain features across diverse conditions and populations.
The lab’s strong presence and impressive achievements at the 2026 ISMRM meeting underscore its leadership in neuroimaging, which will only grow as members continue to innovate in biophysics models and data-processing algorithms to enhance patient care.