Mehmet Akçakaya appointed Jim and Sara Anderson Chair

Professor Mehmet Akçakaya will hold the inaugural Jim and Sara Anderson Chair. Akçakaya leads the Intelligent Medical Imaging and Image Processing (IMAGINE) lab where they develop new image processing technologies for medical applications, particularly magnetic resonance imaging (MRI). The team works at the intersection of artificial intelligence, computational imaging, applied mathematics, and physics. While their areas of application are focused on the brain and the heart, their methods are broadly applicable to many different organs. 

I've been fortunate to hold the inaugural Jim and Sara Anderson Professorship, and now the inaugural Jim and Sara Anderson Chair. These professorship and chair positions give us the much needed support and freedom to explore higher-risk ideas for healthcare technologies that cannot be sponsored by traditional funding mechanisms.

The main line of research in the IMAGINE lab develops new computational approaches to reconstruct high-fidelity images from highly sub-sampled raw magnetic resonance imaging (MRI) data. Their recent efforts on this front rethinks how raw MRI data is processed. Traditionally, image reconstruction precedes tasks such as segmentation or denoising. This year, the team introduced a novel pipeline that applies these tasks directly to the raw frequency-domain data, improving both reconstruction quality and efficiency. In brain imaging, this approach focuses on denoising prior to reconstruction—an essential step for achieving ultra-high spatial resolution without losing fine detail. In cardiac imaging, the focus has been on segmenting and removing static background anatomy (such as the chest wall) directly from the raw data. This simplification allows for higher acceleration and eliminates the need for breath-holds during heart scans.

In parallel, the group is tackling one of MRI’s longstanding limitations: the loud acoustic noise generated during scans, often reaching up to 130 decibels. Since this noise is caused by vibrations of magnetic coils and is directly tied to the scanner’s pulse sequence—known ahead of time—the team is pursuing AI-based predictive noise cancellation. By modeling the MRI system as an acoustic system and incorporating subject-specific calibration scans, the goal is to achieve real-time noise mitigation. With support from the University's Office of Discovery and Translation, the team will build a prototype noise-canceling headset and collect data to train the underlying models.

IMAGINE lab is also working on advances for generative AI models to address challenges of limited-data scenarios in medical imaging. Diffusion models, now considered the state-of-the-art in image processing applications, typically require large datasets for effective training. However, in most medical imaging scenarios, accessible data is inherently limited due to privacy and other issues. Ongoing projects in this front are aimed at improving the training of high-quality diffusion models especially in limited data regimes, and for better control in image generation without retraining.

Akçakaya earned his bachelor's degree in engineering from McGill University in 2005, and his master's and doctoral degrees from Harvard University in 2010.

Jim Anderson earned his bachelor's degree in electrical engineering at the University of Minnesota Twin Cities and is the CEO of Coherent Corporation.