CAREER Goals
Following my postdoctoral training and medical physics imaging residency at Duke University Medical Center, my career goals are to continue (1) translating medical imaging research into clinical practice and (2) educating researchers as well as clinicians on current best practices in medical imaging physics. I chose my current position that consists of clinical, research, and educational components because I believe these responsibilities are synergistic: my clinical time working with and educating physicians and technologists informs my research so that I can address clinically relevant questions and develop translatable answers. In addition to implementing research results into clinical practice, by continuing to conduct research and staying up-to-date with the academic literature, then I will be able to prepare for emerging technologies and anticipate clinical implementation.
Research Accomplishments
In my dissertation research, I built multipurpose and task-specific PET phantoms, authored algorithms for image quality analysis, and conducted computer simulations to parameterize PET image quality as a function of body size. The US adult population's weight is increasing, which degrades PET image quality. My research has shown how PET acquisition protocols can be adapted to account for patient size and how image quality is improved for heavier patients using a recently commercialized technology for patient imaging: time of flight (TOF) PET.
While completing my dissertation research, I built multipurpose and task-specific phantoms, authored image quality analysis algorithms, and simulated data acquisition.
- First, I evaluated the effects that body size has on PET data and image quality using computer simulations. To show a clinical application of the noise versus size relationship, a variable-time PET leg protocol was defined and clinically implemented.
- My primary method for measuring image quality used multiple, small lesion inserts in phantoms. In a single measurement, one small lesion's statistics would suffer from the influence of fluctuations over a few image pixels, but using multiple, approximately independent, lesions increases the statistical power without requiring multiple acquisitions. Such methods were used to characterize reconstruction algorithms (e.g., TOF, penalized likelihood).
- To evaluate image quality as a function of body size, I built a fillable, tapering (FT) phantom that I developed, designed, validated, and characterized using Monte Carlo simulations and phantom acquisitions. The FT phantom was combined with the multisphere image quality analysis to parameterize image quality for TOF PET compared with conventional, non-TOF PET as a function of body size.
Research Goals
My broad research interests are to continuing to evaluate and standardize medical image quality, especially in the context of large, academic hospitals. Regardless of modality, evaluating image quality for a given imaging system requires a battery of tests to characterize the system’s performance: patient set-up, acquisition parameters, data processing / reconstruction, and data transmission. For one imaging system this task is challenging, and a large, academic hospital will have a fleet of imaging equipment that covers a spectrum of modalities, manufacturers, models, software versions, and physical constraints. Although characterizing and optimizing the performance of each individual system is vital, research using a wider perspective must be performed: how does image quality and system performance vary for each modality and across the imaging fleet? This is especially important if there are multiple sites and community clinics.
More specifically, the modality and applications I am primarily interested in are nuclear medicine and molecular imaging image quality metrics, especially in the contexts of a large, academic hospital and among diverse scanner models. Utilization and applications of diagnostic nuclear medicine will continue to increase in oncology, neurology, and cardiology due to several factors including increased access, reimbursement, and validation. Both qualitative and quantitative image-based metrics will continue to be expanded and applied for benchmarking scanner models, ensuring quality control, and optimizing acquisition and reconstruction protocols. Additionally, radiochemistry research is expanding the catalogue of tracers that clinical trials and pharmaceutical corporations are using to assess treatment efficacy and drug delivery.
Image-based metrics will be one of the tools to benchmark scanner models, ensure quality control, and optimize acquisition and reconstruction protocols. In my future research endeavors, I would like to build on my current work by evaluating how best to leverage the improved signal and minimized background that advances in hardware and reconstruction yield. Though a "cleaner image" maybe visually appealing, if it does not improve sensitivity or specificity, then perhaps the improved image quality could be redistributed to make gains elsewhere (e.g., reduced dose, reduce scan time).
I am also interested in extending my research to evaluate and standardize image-based quantitation. New PET radiotracers are creating a burgeoning catalog of choices that clinical trials as well as pharmaceutical corporations are using to assess treatment efficacy and drug delivery. However historically, some parameters in multicenter trial protocols have been underdeveloped: scanner capabilities, image processing, reconstruction, and image analysis, but imaging networks and core labs are becoming more involved.
Last updated: September 1, 2013
