RCRichard E. Carson

Richard E. Carson

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Richard E. Carson has a total of 611 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Evaluation of an image-derived input function and population-based metabolite correction for quantification of synaptic density with [18F]SynVesT-1 and PET in male and female rats

    by Ruth H. Asch, Tutukhanim Balayeva, Zhen W. Zhuang, Krista Fowels, Richard E. Carson, Takuya Toyonaga, Zhengxin Cai

  • Exploratory imaging in SWEDD: Insights from dopaminergic PET, synaptic density PET, and volumetric MRI

    by Faranak Ebrahimian Sadabad, Tommaso Volpi, Sule Tinaz, Praveen Honhar, Lydia Colon, Mahsa Mayeli, Brian Pittman, Mika Naganawa, Mark Dias, Gustavo A. Angarita, et al.

  • Patlak-Guided Self-Supervised Learning for Dynamic PET Denoising

    by Qiong Liu, Xueqi Guo, Yu-Jung Tsai, Jean-Dominique Gallezot, Ming-Kai Chen, Liang Guo, Huidong Xie, Vladimir Panin, Richard E. Carson, Chi Liu

    Quantitative 18F-FDG Patlak Ki images provide physiologically meaningful measures of glucose influx, but they are substantially noisier than SUV images because they are estimated from short-duration dynamic frames. This study aims to improve the quality and quantitative reliability of Ki images by applying a self-supervised deep learning framework for dynamic PET denoising. We propose the Patlak-Guided Self-Supervised Denoising Network (PG-SSDNet), a multi-frame CycleGAN-based model that incorporates a Patlak consistency loss to preserve tracer kinetics across frames. PG-SSDNet was evaluated on 14-frame and 6-frame whole-body 18F-FDG dynamic PET datasets from 27 patients. Performance was compared with a supervised U-Net, a single-frame CycleGAN, and a multi-frame CycleGAN without Patlak guidance. Metrics included mean normalized residual sum of squares (MNRSS) of Patlak fits, liver normalized standard deviation (NSTD), and lesion signal-to-noise ratio (SNR). PG-SSDNet consistently reduced noise while maintaining kinetic fidelity. It achieved the lowest Patlak fitting errors (MNRSS: 0.98% ± 0.14% for 14 frames, 0.85% ± 0.11% for 6 frames), the lowest liver NSTD values (0.105 ± 0.037 for 14 frames, 0.211 ± 0.107 for 6 frames), and the highest lesion SNR across all methods, while preserving accurate mean Ki values. PG-SSDNet effectively denoises 18F-FDG dynamic PET data while preserving Patlak linearity, improving both lesion quantification and image quality. Unlike supervised methods, it requires no paired noisy/clean datasets, making it practical for clinical deployment. Future work will extend PG-SSDNet to other tracers with different kinetic properties and to larger multi-center datasets.

  • Open Access

    Widespread synaptic density loss in schizophrenia follows molecular and network architecture

    by Sidhant Chopra, Patrick D. Worhunsky, Mika Naganawa, Xi-Han Zhang, Ashlea Segal, Loïc Labache, Edwina Orchard, Vanessa Cropley, Stephen Wood, Gustavo A. Angarita, et al.

    Converging neuroimaging, genetic, and post-mortem evidence highlights the fundamental role of synaptic density reductions in schizophrenia pathogenesis. However, the brain-wide spatial pattern of these alterations and the mechanisms underlying this patterning remain to be established. Here, using [ 11 C]UCB-J radiotracer positron emission tomography (PET) imaging in individuals with schizophrenia ( n = 29) and healthy controls ( n = 93), we find a prominent and widespread pattern of lower synaptic density (0.58 < Cohen’s D < 1.47; p FWE < 0.05) in patients. The left hemisphere is substantially more impacted than the right (Cohen’s D = 1.14; p < 0.001), with frontal, temporal, cingulate, thalamic, striatal and hippocampal areas particularly affected. Synaptic density alterations were not spatially aligned with grey matter volume alterations indexed using anatomical Magnetic Resonance Imaging. Lower synaptic density in the left hemisphere is associated with higher normative concentrations of GABA A/BZ , 5HT 2A , mGluR5 and 5HT 1B ( r cca = 0.68; p = 0.022). Simulation-based network diffusion models identified regions that may represent the initial sources of pathology, nominating left inferior frontal areas ( p FWE < 0.05) as potential foci from which synaptic pathology initiates and then propagates to structurally connected and molecularly similar areas. Overall, our findings provide in vivo evidence for widespread synaptic density deficits in schizophrenia that are left-lateralised, independent of grey matter volume alterations, aligned to specific neurochemical systems, and suggest that such synaptic pathology may propagate in a pattern consistent with axonal networks.


  • Iterative Reconstruction Performance on the NeuroEXPLORER: Evaluation with Static and Dynamic Human Imaging

    by Yifan Hu, Jiazhen Zhang, Jean-Dominique Gallezot, Tommaso Volpi, Tianyi Zeng, Takuya Toyonaga, Richard E. Carson

    The NeuroEXPLORER (NX) is a next-generation, brain-dedicated PET scanner offering 10× higher sensitivity compared to the HRRT and < 2 mm spatial resolution. This work evaluated the impact of iterative reconstruction on the quantitative performance of NX using human 18F-SynVesT-1 and 11C-PHNO data with an injection dose of 167.2 ± 13.6 MBq and 275.7 ± 64.8 MBq, respectively. Both 100% count level and 20% down-sampled count level were evaluated. Images were reconstructed with ordered-subsets expectation maximization incorporated PSF and TOF (PSF-TOF-OSEM) using 5 subsets and up to 20 iterations. Quantitative performance of static imaging was evaluated by computing bias and noise from statistically independent list-mode replicates. Simulations incorporating iteration-dependent bias and noise of each dynamic frame were used to evaluate effects on kinetic parameters. In static imaging, 5 subsets × 10 iterations achieved ROI bias < 10% and noise < 1% for 18F-SynVesT-1, and bias < 10% with noise < 10% for 11C-PHNO. In dynamic imaging simulation, with 5 subsets and 15 iterations, kinetic parameter bias remained < 20% and noise < 5%. The results demonstrate quantitative stability and precision of NX across tracers and count levels.


  • Generation of brain PET synaptic density imaging from MRI and FDG-PET using a 3D Multi-stage Residual U-Net

    by Xinyuan Zheng, Patrick Worhunsky, Takuya Toyonaga, Qiong Liu, Xueqi Guo, Yinchi Zhou, Xiongchao Chen, Bo Zhou, Adam P. Mecca, Ming-Kai Chen, et al.

    Synaptic vesicle glycoprotein 2A (SV2A) in human brains is an important biomarker of synaptic loss associated with several disorders, including Alzheimer's disease (AD). However, SV2A tracers, such as [11C]UCB-J, are less available in clinical practice due to constraints that include cost and the need for an onsite cyclotron. As a result, there is interest in generating synthetic SV2A PET images using a more commonly used tracer, or another widely available image modality. We therefore investigated the feasibility of using a deep learning approach to generate synthetic [11C]UCB-J SV2A PET images from T1-weighted (T1w) Magnetic Resonance Imaging (MRI) and/or [18F]FDG PET imaging. In our study, we used paired data from 54 participants for training and evaluation, including 22 healthy controls (HC) and 32 AD patients. We implemented two architectures, 3D UNet and 3D Multi-stage ResUNet (MS ResUNet) for translating either MRI or [18F]FDG PET or both into [11C]UCB-J PET images. Our experiments show that, through progressive learning 3D MS ResUNet achieved less bias compared to a single-stage design. Adopting residual connection, cross-stage feature fusion and supervised attention strategies to our model contributes to better gradient flow, mitigates information loss and enables selective focus, leading to better generation results. Additionally, combining MRI T1w and [18F]FDG PET images provides the best prediction performance, whereas having [18F]FDG PET alone produces better synthetic images than using MRI alone. We also evaluated our model’s performance for 14 typical brain regions. The MS ResUNet achieved less than 5% bias in all regions. Furthermore, the differences between AD and HC in certain regions (e.g., hippocampus and entorhinal) are still significant in generated images (p=0.000). In conclusion, the 3D MS ResUNet was able to generate desirable SV2A PET images that are visually consistent with true images in terms of contrast and tracer distribution, quantitatively with low bias (-0.26% ± 5.56%) and high similarity (R2=0.975), for both AD and HC. Compared with the 3D vanilla UNet (FDG: −2.39% ± 5.55%, MRI: −5.03% ± 13.24%, or both: −1.89% ± 6.04%), the modified design produced more robust SV2A images. The synthetic images can potentially provide additional pathological insights into AD and other neurological disorders without increasing radiation exposure and cost.

  • Open Access

    Optimized reference region and the effect on test-retest reliability and sensitivity to differences between Parkinson’s disease and control groups with [ 11 C]UCB-J

    by Nikkita Khattar, David Matuskey, Jean-Dominique Gallezot, Mika Naganawa, Sophie E Holmes, Faranak Ebrahimian Sadabad, Mahsa Mayeli, Irina Esterlis, Christopher H van Dyck, Adam P Mecca, et al.

    [ 11 C]UCB-J is a radioligand targeting synaptic vesicle glycoprotein 2A, used to image synaptic density. For quantification, a small-volume centrum semiovale area was previously optimized as a [ 11 C]UCB-J reference region (CS2mL); however, due to its small volume, its high variability resulted in reduced reliability. Herein, we evaluated an alternative reference region method to assess longitudinal test-retest reliability and detection of Parkinson’s disease (PD). For estimating distribution volume ratio (DVR), CS2mL and eleven white matter (WM) reference regions (range: 0.5–200 mL) were generated. Same-day and longitudinal test-retest variability (TRV) were assessed (24 healthy subjects (HS); n = 10 same-day and n = 20 longitudinal scans, range: 7–1028 days). Each reference region was used to evaluate the substantia nigra (SN) and caudate DVRs in HS ( n = 25) and PD ( n = 20); 10 mL was the optimal reference region volume, yielding [ 11 C]UCB-J DVR measurements with reduced variability in TRV (same-day: 10 mL: 1.2 ± 5.7%, same-day: CS2mL: −0.9 ± 9.2% longitudinal: 10 mL: 1.5 ± 7.0%, CS2mL: 1.6 ± 11.9%), while maintaining d 0.61 for SN and 0.66 for caudate) compared to CS2mL (0.38 for SN and 0.43 for caudate).


  • Open Access

    High-resolution in utero SV2A PET imaging of the nonhuman primate brain using the NeuroEXPLORER

    by Saloni Mehta, Huili Sun, Nicola Micali, Daniel Holden, Krista Fowles, Samantha Rossano, Jiazhen Zhang, Chao Ma, Paul Han, Wonil Lee, et al.

    The NeuroEXPLORER (NX), a next generation brain-dedicated PET scanner, has demonstrated improved spatial resolution and sensitivity compared with previous PET scanners. In this study, we performed synaptic vesicle glycoprotein 2A (SV2A) PET, an in vivo marker of synaptic density, to evaluate the performance of the NX in both fetal and maternal brains of four pregnant rhesus macaques. Results were compared with the previously acquired PET images using miniEXPLORER and Biograph mCT for fetal and maternal brains (four pregnant monkeys per scanner). The NX demonstrated superior performance compared with the other scanners, providing improved image clarity and higher distribution volume ratios (DVRs), which were significant in the occipital region and putamen of the fetal brain ( p < 0.05, FDR uncorrected), as well as across all regions in the maternal brain ( p < 0.05, FDR corrected). The percentage difference in mean regional DVR between maternal and fetal brains ranged from 36.1% in the amygdala to 78.3% in the occipital lobe in the NX, which may reflect regional differences in the phase of synaptic development. This study highlights the NX as a promising non-invasive tool for investigating in utero synaptic development. The presence of in utero motion emphasizes the necessity of motion correction for reliable PET quantification.


  • Dose Reduction in Dynamic PET Imaging Using Self-Supervised Two-Step Deep Image Prior

    by Andi Li, Mohammad B Syed, Mika Naganawa, David Matuskey, Richard E Carson, Jing Tang

    Reducing the radiotracer dose in dynamic brain PET imaging lowers patient radiation exposure but also degrades the signal-to-noise (SNR) ratio and introduces bias in both dynamic frames and resulting parametric images. These effects compromise the accuracy of quantitative assessments of target proteins. In this study, we propose a self-supervised two-step deep image prior (TS-DIP) method that integrates composite image-based pretraining with individual frame-based fine-tuning to enhance reduced-dose dynamic brain PET imaging. We evaluated the TS-DIP method on 1/10-count dynamic datasets acquired using two tracers with distinct uptake patterns, 11C-UCB-J (targeting synaptic density) and 11C-LSN3172176 (targeting muscarinic cholinergic receptors). Compared with conventional, supervised deep learning, and conditional DIP denoising approaches, the TS-DIP demonstrated a superior tradeoff between noise reduction and bias in both dynamic frames and parametric images. Averaged across all dynamic frames of 11C-UCB-J subjects, the proposed method achieved a 75% reduction in noise with an bias of 2.1%, corresponding to 63% and 60% noise reduction in the K1 (bias, -1.2%) and VT (bias, -3.1%) parametric images, respectively. For 11C-LSN3172176 subjects, the average noise reductions (bias) were 80% (3.4%) in dynamic frames, 62%(-0.38%) in K1 images, and 77% (-3.4%) in VT images. These results highlight the generalizability of the proposed TS-DIP method across tracers with varying activity distributions and its potential to enable accurate, reduced-dose dynamic brain PET imaging without the need for large training datasets.


  • Open Access

    SV2A PET reveals synaptic density loss in experimental autoimmune encephalomyelitis and in a pilot multiple sclerosis study

    by Pou Hong Justin Chia, Takuya Toyonaga, Junchao Tong, Hannah Le, Mark Dias, Amanda J. Boyle, Roger Raymond, Erin E. Longbrake, Yiyun Huang, Richard E. Carson, et al.

    Synaptic loss is increasingly recognized as a key pathological feature in multiple sclerosis (MS), contributing to disease progression and cognitive dysfunction. Synaptic vesicle glycoprotein 2A (SV2A) positron emission tomography (PET) imaging has emerged as a promising tool for quantifying synaptic density in vivo. Here, we used the clinically translatable tracer [ 18 F]SynVesT-1 to comprehensively characterize synaptic density across the brain and spinal cord in the experimental autoimmune encephalomyelitis (EAE) mouse model of MS. In parallel, we performed SV2A PET in patients with MS using the first clinically implemented SV2A radiotracer, [ 11 C]UCB-J, providing cross-species validation of SV2A PET imaging as a biomarker of synaptic pathology. In EAE mice, dynamic [ 18 F]SynVesT-1 PET imaging revealed a significant global reduction in tracer uptake, with nearly 30% decrease in regional distribution volume ( V T ) across all analyzed brain regions ( P < 0.0001). Correspondingly, autoradiography (ARG) corroborated the PET findings, and additional analyses demonstrated reduced SV2A levels in the cervical and lumbar spinal cord. In a clinical PET research study, [ 11 C]UCB-J imaging in MS patients (n = 6) versus age-matched healthy controls (n = 6) showed a 16.4% reduction in global cortical SV2A binding ( P = 0.026), with significant regional reductions of 16 to 26% in several cortical and subcortical subregions. Together, these findings demonstrate that SV2A PET imaging provides a sensitive and quantitative biomarker of synaptic pathology in MS. The consistent reductions in SV2A binding observed in both preclinical and clinical research highlight the role of synaptic degeneration in MS and underscore the utility of SV2A PET imaging in MS research.