Research

Improved Imaging Strategies

We have extensive expertise in pulse sequence design, creating methods for data collection that improve speed and robustness.  We focus much of our work on improving “Signal-to-Noise Ratio” (SNR) efficiency and consistency of image quality.  We design methods for speed by maximizing data acquisition rates, while building sampling strategies that facilitate robust image reconstruction from incomplete data – a method often used for further reductions in scan time.

Much of our current data acquisition technology is built on utilizing and optimizing data acquisition along spiral trajectories, particularly for long acquisition time approaches that improve SNR efficiency.  We have created flexible numerical design strategies as well as simple analytical descriptions for spiral trajectories, along with spiral staircase, sliding arms, and a variety of arm ordering schemes for robust undersampling.  We also explore approaches to limit the audible noise for a better patient experience.

Coupled with spiral acquisition trajectories, we have developed novel approaches to the design of rf pulses, in particular the use of Localized Quadratic (LQ) phase encoding through the use of frequency-swept rf pulses.  These pulses have low power (for reduced heating), low gradient area and moments (for speed and reduced motion sensitivity), and beautifully complement spiral imaging to optimize SNR in pulse sequences that utilize spin echoes.

We design our sequences to be quiet with more consonant (rather than dissonant) acoustic properties, and use minimal rf pulse power to maximize imaging possibilities in patients with implanted medical devices.  We are building the infrastructure to allow our methods to work for body scanning during free-breathing, as opposed to breath-held imaging, which is difficult for many patients.

We also pair our work in data acquisition with computationally efficient methods for robust image reconstruction, often working in the domain of iterative solutions.

Magnetic Field Mapping and Image Deblurring

Robust imaging often depends on a uniform magnetic field, and the path we are pursuing for efficient imaging is particularly sensitive to this requirement.  The properties of human tissues, and metal implants in particular, ensure this requirement is not well met.  Fortunately, one can measure these field changes and mathematically correct for their corruption of the MRI signal during image reconstruction.  This is a critical step to enable accurate, efficient imaging.  We work on robust, rapid ways to measure this field and its dynamics during patient breathing and motion.  We are building computationally efficient algorithms for rapid correction of these effects, along with methods to detect residual error in the final images and apply further corrections as needed.

Novel Approaches to Image Prescription and Scanner Operation

When you adjust your refrigerator, you change a dial that says what temperature you want it to be set at – you don’t adjust the current to the cooling unit, the rate of coolant flow through the coils, the cycle time, the distribution of cooling, etc.  The latter approach exemplifies the current approach to prescribe MR images and operate the MRI scanner – setting physics parameters like Repetition Time, Echo Time, Flip Angle, Echo Train Length – all to get images to look a certain way.  We want to turn this on its head, allowing one to describe the look and feel of the image, and let the scanner make the appropriate adjustments.  This combines technology design of both the pulse sequences (data acquisition logic) as well as the interface.

In practice, technologists must adjust each scan to each patient – depending on their size, ability to hold their breath, the presence of metal, etc.  Current scanners are designed so that some of the parameters that technologists adjust to do their job also unintentionally change the image appearance.  We are working to decouple the parameters and approaches used by technologists from those that affect image contrast and quality, in order to improve image consistency as well as reduce the cognitive load and pressure on the technologist.

Complication Mitigation for Value Improvement

Our general framework for High Value MRI, which encompasses much of what we want to achieve, has three areas of focus.

1. More efficient use, including patient throughput, image quality and consistency, and diagnostic impact.  We want to scan more patients with better and more consistent imaging, focusing on gathering information that will impact treatment and patient outcomes.
2. Reduced barriers to use, including equity of access, scope of use, and patient-specific barriers to care.  Our aspiration for equity of access is to deliver the same high quality care to everyone.  Improving scope of use involves thinking outside of the box, for instance enabling more focused exams in which MRI sessions that are shorter, less expensive, and easier to deliver are used to answer specific questions.  Patient-specific barriers such as claustrophobia and implanted medical devices are also mitigated by our work.
3. Enhanced usability, including safety and user experience of the patient, technologist, and interpreting physician.  We want to deliver the best diagnoses to everyone, and do this in a safe manner that minimizes stress and distraction and maximizes performance for everyone involved.

MRI has a variety of complications, such as a weak and sensitive signal, a complex approach to operation, and a scanner environment that is loud and confining.  These lead to a cascade of additional challenges such as long scan times, technologist distraction, and patient anxiety that interfere with the framework of value described above.  By identifying these root challenges to value, we can build technology and use cases that address them and positively impact multiple expressions of value together.