About

Grace Wang is an Associate Professor of American Studies at UC Davis. She teaches, writes, and researches about Asian American studies, transnational American studies, immigration, race, and music. She is the author of Soundtracks of Asian America: Navigating Race Through Musical Performance (Duke University Press, 2015), a book that explores how Asian Americans use music to construct narratives of self, race, class, and belonging in national and transnational spaces. She is currently at work on a documentary film about timpanist Elayne Jones and a book project analyzing how elite organizations maintain exclusion, through the case study of American orchestras. She received her Ph.D. in American culture from the University of Michigan.

Research topics

• Computer Science
• Artificial Intelligence
• Mathematics
• Statistics
• Nuclear medicine
• Medicine
• Optics
• Biomedical engineering
• Physics
• Radiology

Selected publications

• A predictive model for optimal CT-guided thermal ablation margins in liver malignancies ≤3 cm: a retrospective cohort study

  International Journal of Surgery · 2026-01-13 · 1 citations

  articleOpen access

  BACKGROUND AND AIMS: Standardized quantitative criteria are lacking for the recommended circumferential thermal ablative margin in liver malignancies ≤3 cm. This study aimed to identify predictors of technical success in thermal ablation and develop a predictive model defining optimal ablation margins. METHODS: This retrospective study analyzed 979 patients undergoing computed tomography-guided radiofrequency or microwave ablation for liver malignancies ≤3 cm. Using a radiotherapy treatment planning system, preoperative tumor dimensions [maximum tumor area, preoperative maximum tumor area (PreMTA); tumor volume, preoperative tumor volume (PreTV)] and postoperative ablation characteristics [maximum ablation area, postoperative maximum ablation area (Post-MAA); ablation volume, postoperative ablation volume (PostAV)] were volumetrically quantified. Volume ratio (PostAV/PreTV) and area ratio (PostMAA/PreMTA) were calculated. Univariate and multivariate logistic regression identified predictors of residual tumor. A nomogram was developed and validated using Harrell's C-index, calibration curves, and receiver operating characteristic analysis. Size- and location-specific margin recommendations were derived mathematically. RESULTS: Residual tumor occurred in 91 patients (9.3%). Multivariate analysis identified adjacent large vessel [odd ratio (OR) = 1.97; P = 0.015], larger PreMTA (OR = 1.44; P = 0.014), and lower volume ratio (OR = 0.77; P = 0.009) as independent predictors. For non-perivascular tumors, optimal volume ratios were 4.3, 6.0, and 8.8 (corresponding ΔR = 0.3 cm, 0.8 cm, 1.6 cm) for diameters 1.0 cm, 2.0 cm, and 3.0 cm, respectively. A web-based calculator was implemented (https://fbzl.org/monitor/ablation_model.html). CONCLUSIONS: This study establishes and validates a computational model defining tumor size- and vessel proximity-specific optimal ablation margins for liver malignancies ≤3 cm. The web-accessible model provides evidence-based individualized ablation targets that optimize oncological efficacy while preserving parenchyma, replacing conventional fixed margins.

  PublisherDOI

• Study on the method of selecting objects for investigation of sedimentary radioactive source terms in nuclear power plants in service

  Annals of Nuclear Energy · 2025-08-07

  articleSenior authorCorresponding
  PublisherDOI

• Rapid Low-Energy Gamma-ray Source Localization via Centroid Vector Method

  2025-11-01

  article

  Three-Dimensional position-sensitive CdZnTe (3D-CZT) based Compton cameras offer roomtemperature operation, compactness, high energy resolution, and portability. However, the systems face challenges in performing Compton imaging for low-energy gamma-ray sources. In this work, we apply the Centroid Vector Method (CVM), which depends on the spatial distribution of photopeak events in the detector, to a commercial 3D-CZT based Compton camera MICROARIS MA-S2003 without modifying its system design. We conducted measurements using <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">241</sup>Am and <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">57</sup>Co point-sources and calculated the Position Bias (PB) for each source to evaluate performance. Under low-count conditions, the CVM enables millisecond-level localization of a single low-energy point-source and double point-sources at different energies within 4π Field-of-View (FOV). The algorithm runs on a personal computer and does not require substantial computing resources, offering a cost-effective solution to extend the energy range for gamma-ray source localization using compact Compton cameras, thus enabling potential applications in nuclear security, emergency response, contamination remediation and decommissioning.

  PublisherDOI

• Assessing Scatter Phantom Suitability and Recommending an Appropriate Length for LAFOV and Total-Body PET Based on Human Data

  2025-11-01

  article

  With the growing adoption of long-axial field-of-view (LAFOV) and total-body PET scanners, appropriate system characterization protocols are increasingly important. Conventional NEMA standards recommend a 70 cm scatter phantom, but this does not adequately represent human imaging conditions on LAFOV scanners. This study evaluates the suitability of the conventional NEMA scatter phantom and an extended 175 cm long phantom by comparing them to human data acquired on the uEXPLORER total-body PET scanner over the past six years of operation. Data from numerous clinical patients and research participants were analyzed, to calculate average scatter fraction, human count rates at various activities, and the predominant axial activity distribution in humans. The average human scatter fraction (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$37 \pm 3 \%$</tex>) closely matched those of both the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$70 ~\text{cm}(36.3 \%)$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$175 ~\text{cm}(37.4 \%)$</tex> phantoms. Axial activity distributions indicated that 90 % of total-body activity was contained within an average length of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$112 \pm 7 ~\text{cm}$</tex>, indicating that a phantom longer than 1 meter would better reflect clinical imaging. Count-rate comparisons between human and phantom data further supported the conclusion that an extended phantom configuration is warranted although more analysis is required. Based on the results of this preliminary study we recommend extending the scatter phantom length to at least 1 meter to allow for improved performance evaluations of LAFOV and total-body PET systems.

  PublisherDOI

• Total-Body Parametric Imaging Using Relative Patlak Plot

  Journal of Nuclear Medicine · 2025-02-27 · 7 citations

  articleOpen accessSenior author

  Standard Patlak plot is widely used to describe FDG kinetics for dynamic PET imaging. Whole-body Patlak parametric imaging remains constrained due to the need for a full-time input function. Here, we demonstrate the Relative Patlak (RP) plot, which eliminates the need for the early-time input function, for total-body parametric imaging and its application to clinical 20-min scan acquired in list-mode. We demonstrated that the RP intercept b' is equivalent to a ratio of standardized uptake value relative to the blood, while the RP slope Ki' is equal to the standard Patlak Ki multiplied by a global scaling factor for each subject. One challenge in applying RP to a short scan duration (20 min) is the high noise in parametric images. We applied a deep kernel method for noise reduction. Using the standard Patlak plot as the reference, the RP method was evaluated for lesion quantification, lesion-to-background contrast, and myocardial visualization in total-body parametric imaging with uEXPLORER in 22 human subjects who underwent a 1-h dynamic 18F-FDG scan. The RP method was also applied to the dynamic data regenerated from a clinical standard 20-min scan either at 1-h or 2-h post-injection for two cancer patients. We demonstrated that it is feasible to obtain high-quality parametric images from 20-min dynamic scans using the RP plot with a self-supervised deep-kernel noise reduction strategy. The RP Ki' highly correlated with Ki in lesions and major organs, demonstrating its quantitative potential across subjects. Compared to conventional SUVs, the Ki' images significantly improved lesion contrast and enabled visualization of the myocardium for potential cardiac assessment. The application of RP parametric imaging to two clinical scans also showed similar benefits. Total-body PET with the RP plot is feasible to generate parametric images from the dynamic data of a 20-min clinical scan.

  PublisherOA PDFDOI

• Quantitative Total-Body Imaging of Blood Flow with High-Temporal-Resolution Early Dynamic¹⁸F-FDG PET Kinetic Modeling

  Journal of Nuclear Medicine · 2025-04-30 · 9 citations

  articleOpen accessSenior author

  Past efforts to measure blood flow with the widely available radiotracer ¹⁸F-FDG were limited to tissues with high ¹⁸F-FDG extraction fraction. In this study, we developed an early dynamic ¹⁸F-FDG PET method with high-temporal-resolution (HTR) kinetic modeling to assess total-body blood flow based on deriving the vascular phase of ¹⁸F-FDG transit and conducted a pilot comparison study against a ¹¹C-butanol flow-tracer reference. <b>Methods:</b> The first 2 min of dynamic PET scans were reconstructed at HTR (60 × 1 s/frame, 30 × 2 s/frame) to resolve the rapid passage of the radiotracer through blood vessels. In contrast to existing methods that use blood-to-tissue transport rate as a surrogate of blood flow, our method directly estimated blood flow using a distributed kinetic model (adiabatic approximation to tissue homogeneity [AATH] model). To validate our ¹⁸F-FDG measurements of blood flow against a reference flow-specific radiotracer, we analyzed total-body dynamic PET images of 6 human participants scanned with both ¹⁸F-FDG and ¹¹C-butanol. An additional 34 total-body dynamic ¹⁸F-FDG PET images of healthy participants were analyzed for comparison against published blood-flow ranges. Regional blood flow was estimated across the body, and total-body parametric imaging of blood flow was conducted for visual assessment. AATH and standard compartment model fitting was compared using the Akaike information criterion at different temporal resolutions. <b>Results:</b>¹⁸F-FDG blood flow was in quantitative agreement with flow measured from ¹¹C-butanol across same-subject regional measurements (Pearson correlation coefficient, 0.955; <i>P</i> &lt; 0.001; linear regression slope and intercept, 0.973 and –0.012, respectively), which was visually corroborated by total-body blood-flow parametric imaging. Our method resolved a wide range of blood-flow values across the body in broad agreement with published ranges (e.g., healthy cohort values of 0.51 ± 0.12 mL/min/cm³ in the cerebral cortex and 2.03 ± 0.64 mL/min/cm³ in the lungs). HTR (1–2 s/frame) was required for AATH modeling. <b>Conclusion:</b> Total-body blood-flow imaging was feasible using early dynamic ¹⁸F-FDG PET with HTR kinetic modeling. This method may be combined with standard ¹⁸F-FDG PET methods to enable efficient single-tracer multiparametric flow-metabolism imaging, with numerous research and clinical applications in oncology, cardiovascular disease, pain medicine, and neuroscience.

  PublisherOA PDFDOI

• Kinetic modeling of 18F-PI-2620 binding in the brain using an image-derived input function with total-body PET

  EJNMMI Research · 2025-05-30 · 1 citations

  articleOpen access

  Abstract Background Accurate quantification of tau binding from 18 F-PI-2620 PET requires kinetic modeling and an input function. We aimed to implement a non-invasive Image-derived input function (IDIF) using the state-of-the-art total-body uEXPLORER PET/CT scanner to quantify tau binding and tracer delivery rate from 18 F-PI-2620 in the brain. Additionally, we investigated the impact of scan duration on the quantification of kinetic parameters. Results 18 F-PI-2620 total-body PET dynamic (90 min) data from 15 elderly (66–92 years) participants were acquired. Time-activity curves were obtained from grey matter regions of interest (ROIs) known to be affected in Alzheimer’s disease, including the medial temporal lobe, posterior cingulate, and lateral parietal cortex. These curves were fitted to the two-tissue compartmental model (2TCM) using a subject-specific IDIF (plasma and metabolite corrected) derived from the descending aorta. ROI-specific kinetic parameters were estimated for different scan durations ranging from 10 to 90 min. The parameters included blood fraction volume (v b ), rate constants (K 1 , k 2 , k 3 , k 4 ), total distribution volume (V T ), distribution volume ratio (DVR), and tracer arrival delay. Logan graphical analysis was also used to estimate V T and compared with 2TCM. Differences in kinetic parameters were observed between ROIs, including significant reduction in tracer delivery rate (K 1 ) in the medial temporal lobe (q &lt; 0.001). All kinetic parameters remained relatively stable (compared to parameters quantified with full 90-minute data) after the 60-minute scan window across all ROIs ( r ≥ 0.89; p &lt; 0.001), with K 1 showing high stability after 30 min of scan duration ( r ≥ 0.92; p &lt; 0.001). Excellent correlation was observed between V T estimated using 2TCM and Logan plot analysis ( r ≥ 0.96; p &lt; 0.001). Conclusions This study demonstrated the utility of IDIF from a lager blood pool, derived using the total-body PET in quantifying 18 F-PI-2620 kinetics in the brain. Our findings suggest that a 60-minute scan window may be required for the reliable quantification of kinetic parameters using IDIF, whereas a 30-minute scan time may be sufficient for the quantification of K 1 .

  PublisherOA PDFDOI

• Maximum Likelihood Reconstruction of Attenuation and Activity (MLAA) in SPECT for Improved Attenuation Correction: Simulation, Phantom, and Patient Data Validation

  2025-11-01

  article

  Purpose: Quantitative SPECT-CT, originally developed for radionuclide therapy dosimetry, relies on accurate attenuation maps. As attenuation is energy-dependent, CT-derived data must be converted to the radionuclide's photopeak energy (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mu$</tex>-map)[1], typically via bilinear scaling, which can introduce bias. To address this, we apply the MLAA algorithm, which enhances <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mu$</tex>-map accuracy by incorporating object-specific information from the photopeak emission sinogram.

  PublisherDOI

• Correction to: Feasibility of PET‑enabled dual‑energy CT imaging: First physical phantom and initial patient study results

  European Journal of Nuclear Medicine and Molecular Imaging · 2025-01-08

  erratumOpen accessSenior author
  PublisherOA PDFDOI

• Abstract 4369925: Brain and Cardiovascular Connection in Recovery: Coupling Neurovascular Regulation to Vascular and Non-cardiac Inflammation in the Individual Patient Post Acute Myocardial Infarction

  Circulation · 2025-11-03

  articleSenior author

  Background: How the human body coordinates cerebral blood flow (CBF) and multi-organ inflammation during recovery of an acute myocardial infraction (AMI) is difficult to study and poorly understood. This is in part due to a lack of noninvasive measuring techniques for both blood flow (BF) and total-body (TB) inflammation per organ in treated survivors. Conventional PET imaging can either resolve organ-specific BF or tissue inflammation with a singular tracer injection but not both. Hypothesis: We hypothesize that early kinetics and delayed static images from a single tracer injection for PET scans will simultaneously evaluate and connect, for the first time, CBF abnormalities and remote vascular (i.e. aorta) and/or solid organ inflammation at a singular time point. Methods: Here, we expand our prior work on multiparametric TB-PET with high-temporal resolution dynamic imaging (1-2 s/frame for the first ~2 mins of scanning) for CBF modeling with a no-flow radiotracer (i.e. 18 F-FDG) based on early vascular transit time (VTT) in the gray and white matter, brainstem, and cerebellum. The whole aorta, by target–to–blood pool ratio (TBR) at 40-60 min, and solid organs, by standardized uptake value (SUV) at 60-90 min, were evaluated for inflammation based on glucose uptake. Result: Eleven revascularized survivors (~10 days post event) and 22 non-AMI subjects were studied. CBF was primarily reduced, when compared to controls, in subcortical gray matter (0.353 vs. 0.434 mL/min/cm 3 , p=0.0121) with a lesser reduction trend in cortical gray matter (0.414 vs. 0.449, p= 0.0677). This was accompanied by a gray matter subcortical increase in mean VTT (6 vs. 4.3 sec, p=0.0121) and to a lesser extend a cortical increase (5.5 vs. 4.5 sec, p=0.0253). The TBR and SUVs of extra-cardiac, non-cerebral organs were increased 1.9 vs. 1.23, p&lt;0.0001 in the whole aorta, 2.9 vs. 2.2 p=0.0435 in the bone marrow and 3.1 vs. 2.1 p=0.0015 in the spleen respectively. Conclusion: When simultaneously evaluating extra-cardiac organs in survivors, we found a widespread pattern of multi-organ inflammation and a restricted-to-regional hypoperfusion of subcortical &gt; cortical gray matter using dynamic and static total-body 18 F-FDG PET/CT imaging per organ, per patient and with a singular non-flow tracer injection. Hence, future longitudinal PET imaging on AMI survivors as shown offers a unique opportunity to unravel the complex process of recovery and the multiorgan contribution to resilience post MI.

  PublisherDOI

Recent grants

• Cardiac CT Deblooming

  NIH · $3.8M · 2020–2026

• High spatial resolution dedicated head and neck PET system based on cadmium zinc telluride detectors

  NIH · $2.6M · 2018–2024

• Single-tracer Multiparametric PET Imaging

  NIH · $2.4M · 2022–2026

• PET-enabled Dual-energy CT

  NIH · $628k · 2019–2022

• NIH Grant R21HL131385

  NIH · $431k · 2019

Frequent coauthors

• Ramsey D. Badawi

  53 shared

• Jinyi Qi

  University of California, Davis

  52 shared

• Simon R. Cherry

  University of California, Davis

  43 shared

• Wendong Xu

  Huashan Hospital

  30 shared

• Fei Gao

  Sun Yat-sen University Cancer Center

  30 shared

• Benjamin A. Spencer

  University of California, Davis

  29 shared

• Ai-Ping Yu

  Shanghai Jiao Tong University

  26 shared

• Fujun Zhang

  Sun Yat-sen University Cancer Center

  26 shared