Sarah Zou

All Publications

• Comparison of deep learning and particle smoother EM methods for estimation of Rb-82 myocardial perfusion PET kinetic parameters. Medical physics Chin, M., Zou, S. J., Chinn, G., Levin, C. S. 2026; 53 (6): e70512

  Abstract

  Positron emission tomography (PET) enables quantification of dynamic physiological processes through time-resolved imaging. In 82 Rb $^{82}{\rm Rb}$ myocardial perfusion PET, kinetic compartment modeling is used to estimate physiological parameters and derive myocardial blood flow. However, conventional nonlinear least squares (NLLS) estimation is sensitive to model misspecification when not all parameters can be reliably estimated and must instead be fixed or initialized using population averages, which can degrade accuracy.This work develops and evaluates two alternative kinetic analysis approaches for 82 Rb $^{82}{\rm Rb}$ PET: a particle smoother-based Expectation-Maximization method (PSEM) and a convolutional neural network (CNN).Both methods were evaluated using simulated 82 Rb $^{82}{\rm Rb}$ dynamic myocardial perfusion studies and compared against NLLS and a Kalman-smoother-based Expectation-Maximization (KEM) algorithm across multiple frame durations and noise levels.Across 2-10 s frames, the CNN achieved the lowest relative errors for all parameters ( F $F$ : 8.78%-4.98%, k 3 $k_3$ : 26.05%-25.50%, k 4 $k_4$ : 34.34%-22.76%), significantly outperforming NLLS, KEM, and PSEM (Holm-adjusted p < 10 - 15 $p < 10^{-15}$ at 1.0 × $\times$ noise, 2-s frames), although performance degraded under out-of-distribution input-function conditions.Overall, the CNN provided the most accurate and robust in-distribution kinetic parameter estimates across frame durations. In contrast, PSEM exhibited parameter-dependent behavior, improving k 3 $k_3$ estimation while underperforming for F $F$ , suggesting that further methodological refinement is needed.

  View details for DOI 10.1002/mp.70512

  View details for PubMedID 42253226

• Maximum likelihood estimation yields accurate line-of-response assignment for positron + prompt gamma ray events in multiplexed PET (mPET). Biomedical physics & engineering express Zou, S. J., Chinn, G., Ullah, M. N., Levin, C. S. 2026

  Abstract

  For accurate disease characterization using positron emission tomography (PET), it is desirable to image multiple radiotracers in a single scan. Conventional PET methods cannot do this due to the indistinguishable annihilation photons produced by different radiotracers. One approach is to label one radiotracer with a positron+prompt-gamma ($\beta^+\!\!-\!\!\gamma$) isotope producing triple coincidences, and another with a pure positron-emitting ($\beta^+$) isotope producing double coincidences. However, $\beta^+\!\!-\!\!\gamma$ emitters present challenges in correctly identifying the two annihilation photons, or equivalently, assigning the correct line-of-response (LOR) to triple-photon coincidence events. Here, we propose a maximum likelihood estimation (MLE) framework leveraging spatial, timing, and energy information to determine the most probable LOR. Simulation studies validated the method: simulations showed over 96\% and 94\% accuracy for LOR assignment of $\beta^+\!\!-\!\!\gamma$ emitters $^{22}$Na and $^{124}$I point sources, respectively. Furthermore, simulated phantom imaging of $^{22}$Na or $^{124}$I distributions alongside a $\beta^+$ emitter demonstrated that MLE LOR assignment achieved comparable image quality-measured by contrast recovery coefficient (CRC) and cross-talk ratio (XR)-to benchmark methods, where the prompt gamma was identified using an energy threshold ($\geq 650$ keV) for $^{22}$Na and as the highest-energy photon for $^{124}$I.

  View details for DOI 10.1088/2057-1976/ae6e44

  View details for PubMedID 42140224

• Triplexed PET: Methods to Image Three Distinct Isotopes in a Single PET Imaging Session Zou, S. J., Ullah, M. N., Innes, D., Shah, Y., Chinn, G., Houson, H., Lapi, S., Levin, C. SOC NUCLEAR MEDICINE INC. 2025

  View details for Web of Science ID 001538700900003

• QUANTITATIVE IMAGING OF <SUP>55</SUP>CO AND <SUP>18</SUP>F-LABELED TRACERS IN A SINGLE "MULTIPLEXED" PET IMAGING SESSION Zou, S. J., Lim, I., Foster, J. W., Chinn, G., Houson, H. A., Lapi, S. E., Rao, J., Levin, C. S., IEEE IEEE. 2025

  View details for DOI 10.1109/ISBI60581.2025.10980731

  View details for Web of Science ID 001546451000081

• PETcoil: first results from a second-generation RF-penetrable TOF-PET brain insert for simultaneous PET/MRI. Physics in medicine and biology Dong, Q., Ullah, M. N., Innes, D. R., Watkins, R. D., Chang, C., Zou, S. J., Groll, A., Sacco, I., Chinn, G., Levin, C. S. 2024

  Abstract

  Simultaneous PET/MRI provides concurrent information about anatomic, functional, and molecular changes in disease. We are developing a second generation MR-compatible RF-penetrable TOF-PET insert. The insert has a smaller scintillation crystal size and ring diameter compared to clinical whole-body PET scanners, resulting in higher spatial resolution and sensitivity. This paper reports the initial system performance of this full-ring PET insert. The global photopeak energy resolution and global coincidence time resolution, 11.74 ± 0.03 % FWHM and 238.1 ± 0.5 ps FWHM, respectively, are preserved as we scaled up the system to a full ring comprising 12,288 LYSO-SiPM channels. Throughout a ten-hour experiment, the system performance remained stable, exhibiting a less than 1% change in all measured parameters. In a resolution phantom study, the system successfully resolved all 2.8 mm diameter rods, achieving an average VPR of 0.28 ± 0.08 without TOF and 0.24 ± 0.07 with TOF applied. Moreover, the implementation of TOF in the Hoffman phantom study also enhanced image quality. Initial MR compatibility studies of the full PET ring were performed with it unpowered as a milestone to focus on looking for material and geometry-related artifacts. During all MR studies, the MR body coil functioned as both the transmit and receive coil, and no observable artifacts were detected. As expected, using the body coil also as the RF receiver, MR image signal-to-noise ratio exhibited degradation (∼30%), so we are developing a high quality receive-only coil that resides inside the PET ring.

  View details for DOI 10.1088/1361-6560/ad7221

  View details for PubMedID 39168156

• Pecan Shell-Derived Activated Carbon for High-Electrochemical Performance Supercapacitor Electrode. Materials (Basel, Switzerland) Zou, S. J., Patel, M. D., Smith, L. M., Cha, E., Shi, S. Q., Choi, W. 2024; 17 (13)

  Abstract

  Carbon nanomaterials-based electric double-layer capacitors (EDLCs) are reliable and appealing energy-storage systems offering high power density and long cycling stability. However, these energy storage devices are plagued with critical shortcomings, such as low specific capacitance, inefficient physical/chemical activation process, and self-discharge of electrode materials, hindering their future application. In this work, we use a self-activation process, an environmentally benign and low-cost process, to produce high-performance activated carbon (AC). Novel activated carbon from pecan shells (PS) was successfully synthesized through a single-step self-activation process, which combines the carbonization and activation processes. The as-synthesized pecan shell-derived activated carbon (PSAC) provides a high-porosity, low-resistance, and ordered pore structure with a specific pore volume of 0.744 cm3/g and BET surface area of 1554 m2/g. The supercapacitors fabricated from PSAC demonstrate a specific capacitance of 269 F/g at 2 A/g, excellent cycling stability over 15,000 cycles, and energy and power density of 37.4 Wh/kg and of 2.1 kW/kg, respectively. It is believed that the high-efficiency PSAC synthesized from the novel self-activation method could provide a practical route to environmentally friendly and easily scalable supercapacitors.

  View details for DOI 10.3390/ma17133091

  View details for PubMedID 38998174