Clinical applications of fast, quantitative MR fingerprinting

Dan Ma is an assistant professor in the Department of Biomedical Engineering and the School of Medicine at Case Western Reserve University. At SPIE Medical Imaging 2023, Ma will discuss how combining the precision and sensitivity of magnetic resonance fingerprinting (MRF) with advanced image analysis techniques could lead to a unified quantitative imaging tool for precision cancer imaging and treatment planning.

How would you describe “magnetic resonance fingerprinting (MRF)?”
Most clinical magnetic resonance imaging (MRI) scans are inherently qualitative measurements, which only provide relative intensity contrasts between different tissues. Images from these qualitative MRI scans are hard to interpret objectively, difficult to reproduce, and can vary — even within the same hospital — based on the type of scanners. Although fully quantitative MRI has long been the goal of research in MR, clinical translation of these methods faces challenges due to long scan time and low robustness.

MR Fingerprinting is a scan that generates multiple quantitative tissue maps from a single scan within a feasible clinical scan timeframe. Specifically, MRF scans apply various radiofrequency pulses and acquisition timings to make the acquired signal different from different tissue types. Then a ‘dictionary’ is built that stores signals simulated from a wide range of tissue properties. By using a pattern recognition algorithm, the acquired signals are ‘matched’ to the signals in the dictionary. Once the match is identified, the underlying tissue properties that determine the signal can be extracted and translated into quantitative maps.

What are some of the clinical applications for MRF?
MRF is a general framework that offers high flexibility in acquisition design, so there are various designs of MRF scans, providing different quantitative maps for different clinical purposes.

For example, the MRF scan that provides T1 and T2 relaxation maps has been used in brain tumors, epilepsy, Alzheimer’s disease, and multiple sclerosis studies. Apart from neuro, this MRF scan has been adapted to study prostate cancer, breast cancer, and cardiac diseases.

Other MRF scan designs have been proposed to generate T2-star (T2*) perfusion, diffusion, or chemical exchange maps in a clinically feasible time. These scans have also been used to study stroke, neurodegenerative diseases, brain tumors, liver diseases, and so on.

What are some of the current projects at the Ma Laboratory (Quantitative MR Imaging Lab) that you’re most excited about? 
Ongoing projects include a long-term collaboration with the Epilepsy Center at Cleveland Clinic. The project is to characterize epilepsy lesions and detect subtle focal lesions of epilepsy patients using MRF. We are expanding this study to multiple sites, and I look forward to this large-scale study.

Recently I received an academic industrial partnership grant from NIH. The collaboration with Siemens Healthineers, University Hospitals Cleveland Medical Center, and the University of Pennsylvania will accelerate the technical and clinical translation of MRF, with a clear goal of commercialization and clinical workflow integration.

Another exciting project is ‘making invisible to visible,’ which is an international collaboration with Cardiff University, Leeds University, and University College London in the UK. We will integrate tissue imaging, tissue modeling, MR scans, and AI to quantify tissue microstructure in the human cortex.

What do you want attendees to learn from your talk?
First, quantitative MR scans can address many limitations of the current clinical weighted MR scans. Quantitative MR scans can be efficient and have great potential in clinical translation.

Second, MR Fingerprinting is a very flexible MR framework. Various designs allow you to generate different tissue property maps. Therefore, there are plenty of opportunities in new MR scan design and scan optimization.

Finally, there are plenty of opportunities in quantitative image analysis. Joint optimization of image acquisition and analysis may provide better solutions for clinical applications.