Siqi Li (Ph.D., Northeastern University, China) is an Assistant Project Scientist at Radiology Department, University of California, Davis.
Embedded Files
Research Interests
Research Interests
Multiparametric PET imaging with AI, e.g., Tomographic imaging enhancement (PET, CT); Kinetic modeling; Total-body imaging
Medical image analysis, e.g., Computer-aided segmentation and detection based on AI technology
Deep learning, e.g., Interpretable deep learning and visualization
Clinical investigation, e.g., Combination with deep learning to improve clinical application of medical imaging
Current Projects
Current Projects
- High-temporal resolution PET imaging
Current clinical PET can provide an effective temporal resolution of 10 seconds but with compromised image quality. We are aiming to explore higher temporal resolution (HTR) imaging, for example, 1-2 seconds with current clinical PET scanners and sub-second with high-sensitivity PET scanners. Here a data-driven deep kernel method and neural kernelized expectation-maximization (neural KEM) method are developed for improving PET reconstruction.
Current clinical PET can provide an effective temporal resolution of 10 seconds but with compromised image quality. We are aiming to explore higher temporal resolution (HTR) imaging, for example, 1-2 seconds with current clinical PET scanners and sub-second with high-sensitivity PET scanners. Here a data-driven deep kernel method and neural kernelized expectation-maximization (neural KEM) method are developed for improving PET reconstruction.
Related publications:
S. Li, K. Gong, R. D. Badawi, E. J. Kim, J. Qi and G. Wang, "Neural KEM: A Kernel Method With Deep Coefficient Prior for PET Image Reconstruction," in IEEE Transactions on Medical Imaging, 42(3): 785-796, 2023. (Open code on GitHub)
S. Li and G. Wang, "Deep Kernel Representation for Image Reconstruction in PET," in IEEE Transactions on Medical Imaging, 41(11): 3029-3038, 2022. (Open code on GitHub)
- PET-enabled dual-energy CT imaging
PET-enabled dual-energy CT concept has been proposed to allow DECT imaging on PET/CT scanners without using a second X-ray CT scan. This is achieved by combining a low-energy X-ray CT image with a 511 keV attenuation image (GCT) reconstructed from time-of-flight PET emission data. The maximum-likelihood attenuation and activity (MLAA) algorithm can obtain the GCT image but suffers from heavy noise. We are developing deep learning-based kernel methods for improving GCT image reconstruction, which in turn improves material decomposition.
PET-enabled dual-energy CT concept has been proposed to allow DECT imaging on PET/CT scanners without using a second X-ray CT scan. This is achieved by combining a low-energy X-ray CT image with a 511 keV attenuation image (GCT) reconstructed from time-of-flight PET emission data. The maximum-likelihood attenuation and activity (MLAA) algorithm can obtain the GCT image but suffers from heavy noise. We are developing deep learning-based kernel methods for improving GCT image reconstruction, which in turn improves material decomposition.
Related publications:
S. Li and G. Wang, "Modified Kernel MLAA Using Autoencoder for PET-enabled Dual-energy CT," in Philosophical Transactions of the Royal Society A, 379: 20200204, 2021. (Open code on GitHub)
S. Li and G. Wang, "Neural MLAA for PET-enabled Dual-Energy CT Imaging," SPIE Medical Imaging 2021: Physics of Medical Imaging, 115951G (15 February 2021).
- Clinical investigation with dynamic PET imaging
Dynamic PET imaging can monitor the 4D spatiotemporal distribution of a radiotracer in the living body. Parametric imaging from 4D dynamic PET data can provide more meaningful physiological information using voxel-wise tracer kinetic modeling. We are investigating the potential clinical applications using dynamic PET imaging with tracer kinetic modeling, looking forward to providing valuable information (e.g., image biomarkers) for improving clinical diagnosis.
Dynamic PET imaging can monitor the 4D spatiotemporal distribution of a radiotracer in the living body. Parametric imaging from 4D dynamic PET data can provide more meaningful physiological information using voxel-wise tracer kinetic modeling. We are investigating the potential clinical applications using dynamic PET imaging with tracer kinetic modeling, looking forward to providing valuable information (e.g., image biomarkers) for improving clinical diagnosis.
Related publications:
Related publications:
S. Li, Y Wang, B. A. Spencer, H. Hunt, T. Seibert, L. Nardo, S. R. Cherry, R. D. Badawi and G. Wang, "Dual-energy CT Bone-fraction Correction for Total-body PET Kinetic Quantification of Bone Marrow," SNMMI Annual Meeting 2022, accepted for poster presentation.
Publications
Publications
Selected Journal Papers
Selected Journal Papers
S. Li, Y. Zhu, B. A. Spencer and G. Wang, "Single-Subject Deep-Learning Image Reconstruction With a Neural Optimization Transfer Algorithm for PET-Enabled Dual-Energy CT Imaging," in IEEE Transactions on Image Processing, vol. 33, pp. 4075-4089, 2024.
S. Li, K. Gong, R. D. Badawi, E. J. Kim, J. Qi and G. Wang, "Neural KEM: A Kernel Method With Deep Coefficient Prior for PET Image Reconstruction," in IEEE Transactions on Medical Imaging, 42(3): 785-796, 2023. (Open code on GitHub)
S. Li and G. Wang, "Deep Kernel Representation for Image Reconstruction in PET," in IEEE Transactions on Medical Imaging, 41(11): 3029-3038, 2022. (Open code on GitHub)
S. Li and G. Wang, "Modified Kernel MLAA Using Autoencoder for PET-enabled Dual-energy CT," in Philosophical Transactions of the Royal Society A, 379: 20200204, 2021. (Open code on GitHub)
S. Li, H. Jiang, H. Li and Y. -D. Yao, "AW-SDRLSE: Adaptive Weighting and Scalable Distance Regularized Level Set Evolution for Lymphoma Segmentation on PET Images," in IEEE Journal of Biomedical and Health Informatics, 25(4): 1173-1184, 2021.
S. Li, H. Jiang, J. Bai, Y. Liu and Y. -D. Yao, "Stacked Sparse Autoencoder and Case-based Postprocessing Method for Nucleus Detection," in Neurocomputing, 359(24): 494-508, 2019.
S. Li, H. Jiang, Y. -D. Yao, W. Pang, Q. Sun and L. Kuang, "Structure Convolutional Extreme Learning Machine and Case-based Shape Template for HCC Nucleus Segmentation," in Neurocomputing, 312(27): 9-26, 2018. (SC-ELM code on GitHub)
S. Li, H. Jiang, Z. Wang and Y. -D. Yao, "An Effective Computer Aided Diagnosis Model for Pancreas Cancer on PET/CT Images," in Computer Methods and Programs in Biomedicine, 165: 205-214, 2018.
S. Li, H. Jiang, Y. -D. Yao and B. Yang, "Organ Location Determination and Contour Sparse Representation for Multiorgan Segmentation," in IEEE Journal of Biomedical and Health Informatics, 22(3): 852-861, 2018.
S. Li, H. Jiang and W. Pang, "Joint Multiple Fully Connected Convolutional Neural Network with Extreme Learning Machine for Hepatocellular Carcinoma Nuclei Grading," in Computers in Biology and Medicine, 84(1): 156-167, 2017.
Selected Conference Papers
Selected Conference Papers
S. Li, and G Wang, "Learning of Deep Kernels with Pairwise Attention for PET Image Reconstruction," IEEE NSS-MIC 20202, accepted for oral presentation.
S. Li, Y Wang, B. A. Spencer, H. Hunt, T. Seibert, L. Nardo, S. R. Cherry, R. D. Badawi and G. Wang, "Dual-energy CT Bone-fraction Correction for Total-body PET Kinetic Quantification of Bone Marrow," SNMMI Annual Meeting 2022, accepted for poster presentation.
S. Li, and G. Wang, "A Deep Kernel Method for PET Image Reconstruction," SPIE Medical Imaging 2022: Image Processing. Vol. 12032. SPIE, 2022.
S. Li, K. Gong, J. Qi and G. Wang, "Neural KEM for PET Image Reconstruction," IEEE NSS-MIC 20201, accepted for oral presentation.
S. Li and G. Wang, "Kernel MLAA Using Autoencoder for PET-enabled Dual-Energy CT," 16th Virtual International Meeting on Fully 3D Image Reconstruction, Leuven Belgium, July 2021.
S. Li and G. Wang, "Neural MLAA for PET-enabled Dual-Energy CT Imaging," SPIE Medical Imaging 2021: Physics of Medical Imaging, 115951G (15 February 2021).
S. Li and G. Wang, "Low-dose CT Image Denoising using Parallel-clone Networks," IEEE NSS-MIC 2020, accepted for oral presentation. [Preprint: arXiv:2005.06724. May 2020]
More publications can be found on Google Scholar
Contact Me
Contact Me
Email: sqlli@ucdavis.edu
Page updated
Google Sites
Report abuse