Suma Anand

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2024-230

December 20, 2024

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2024/EECS-2024-230.pdf

Magnetic Resonance Imaging (MRI) is a powerful noninvasive imaging modality that can visualize soft tissue structures deep in the body without harmful ionizing radiation. MRI has numerous uses in medicine and neuroscience, such as for high-resolution structural imaging of the brain, imaging of moving organs like the heart and the liver, and integration with other modalities such as MR-guided radiation therapy. However, MRI is fundamentally slow, and motion inevitably arises during scans that can take seconds to minutes. In fact, patient motion is the most common unanticipated event in a clinical MRI exam. Motion results in degraded image quality, repeated exams, and increased costs. Suppressing motion is insufficient; instead, a better strategy is to estimate and correct motion by measuring it with external sensors.

In this dissertation, I introduce Beat Pilot Tone (BPT), a new sensor that has the potential to turn any MRI scanner into a general-purpose Radio Frequency (RF) motion monitoring system. BPT involves transmitting RF tones at arbitrary frequencies that create motion-modulated standing wave patterns. These are sensed by the same coil array used for MR imaging and received via intermodulation in the coil preamplifiers.

I explore the origins of the BPT signal in electromagnetic field simulations that are validated by experimental measurements. In extensive volunteer experiments, I use BPT to sense different motion types, such as respiratory, bulk, cardiac, and head motion. I demonstrate that it can detect small body movements (e.g. vibrations) and show that it is up to seven times more modulated than a state-of-the-art motion sensor known as Pilot Tone (PT). Furthermore, I show that by using multiple-input multiple-output (MIMO) BPTs, we can distinguish between different head movements (nodding "yes" and shaking "no").

In later chapters, I develop the implementation and application of MIMO-BPT for retrospective head motion correction. I describe and characterize the hardware system for acquiring simultaneous MIMO-BPT and PT signals. Using this system, I introduce the first-ever dataset of head motion during brain scans with MIMO-BPT and PT, collected on 13 healthy volunteers. Preliminary results from this dataset suggest that BPT may outperform PT in correcting head motion quantitatively. I conclude by suggesting future improvements and applications of BPT.

Advisors: Michael Lustig


BibTeX citation:

@phdthesis{Anand:EECS-2024-230,
    Author= {Anand, Suma},
    Title= {Motion Sensing and Correction in Magnetic Resonance Imaging with Radio Frequency Signals},
    School= {EECS Department, University of California, Berkeley},
    Year= {2024},
    Month= {Dec},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2024/EECS-2024-230.html},
    Number= {UCB/EECS-2024-230},
    Abstract= {Magnetic Resonance Imaging (MRI) is a powerful noninvasive imaging modality that can visualize soft tissue structures deep in the body without harmful ionizing radiation. MRI has numerous uses in medicine and neuroscience, such as for high-resolution structural imaging of the brain, imaging of moving organs like the heart and the liver, and integration with other modalities such as MR-guided radiation therapy. However, MRI is fundamentally slow, and motion inevitably arises during scans that can take seconds to minutes. In fact, patient motion is the most common unanticipated event in a clinical MRI exam. Motion results in degraded image quality, repeated exams, and increased costs. Suppressing motion is insufficient; instead, a better strategy is to estimate and correct motion by measuring it with external sensors.

In this dissertation, I introduce Beat Pilot Tone (BPT), a new sensor that has the potential to turn any MRI scanner into a general-purpose Radio Frequency (RF) motion monitoring system. BPT involves transmitting RF tones at arbitrary frequencies that create motion-modulated standing wave patterns. These are sensed by the same coil array used for MR imaging and received via intermodulation in the coil preamplifiers.

I explore the origins of the BPT signal in electromagnetic field simulations that are validated by experimental measurements. In extensive volunteer experiments, I use BPT to sense different motion types, such as respiratory, bulk, cardiac, and head motion. I demonstrate that it can detect small body movements (e.g. vibrations) and show that it is up to seven times more modulated than a state-of-the-art motion sensor known as Pilot Tone (PT). Furthermore, I show that by using multiple-input multiple-output (MIMO) BPTs, we can distinguish between different head movements (nodding "yes" and shaking "no"). 

In later chapters, I develop the implementation and application of MIMO-BPT for retrospective head motion correction. I describe and characterize the hardware system for acquiring simultaneous MIMO-BPT and PT signals. Using this system, I introduce the first-ever dataset of head motion during brain scans with MIMO-BPT and PT, collected on 13 healthy volunteers. Preliminary results from this dataset suggest that BPT may outperform PT in correcting head motion quantitatively. I conclude by suggesting future improvements and applications of BPT.},
}

EndNote citation:

%0 Thesis
%A Anand, Suma 
%T Motion Sensing and Correction in Magnetic Resonance Imaging with Radio Frequency Signals
%I EECS Department, University of California, Berkeley
%D 2024
%8 December 20
%@ UCB/EECS-2024-230
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2024/EECS-2024-230.html
%F Anand:EECS-2024-230