Article
Mathematics, Applied
Zeljko Kereta, Robert Twyman, Simon Arridge, Kris Thielemans, Bangti Jin
Summary: This study investigates two classes of algorithms for accelerating OSEM based on variance reduction, resulting in significantly improved efficiency and accuracy for penalized PET reconstructions.
Article
Computer Science, Interdisciplinary Applications
Siqi Li, Kuang Gong, Ramsey. D. D. Badawi, Edward J. Kim, Jinyi Qi, Guobao Wang
Summary: This paper proposes an implicit regularization method for kernel methods in image reconstruction of low-count PET data. By using a convolutional neural network to represent the kernel coefficient image in the PET forward model, the proposed method achieves better performance than existing methods. The neural KEM algorithm, which combines the KEM step and deep learning step, is demonstrated to be effective in both computer simulations and real patient data.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Engineering, Biomedical
Seung Kwan Kang, Jae Sung Lee
Summary: This study proposes a Bowsher prior based on the l(1)-norm and an iteratively reweighting scheme for Anatomy-Guided regularized PET image reconstruction, which overcomes the limitations of the original Bowsher method and improves small lesion detection and contrast enhancement.
PHYSICS IN MEDICINE AND BIOLOGY
(2021)
Article
Computer Science, Interdisciplinary Applications
Kibo Ote, Fumio Hashimoto, Yuya Onishi, Takashi Isobe, Yasuomi Ouchi
Summary: This study proposes a novel list-mode PET image reconstruction method using an unsupervised CNN called deep image prior (DIP). The proposed LM-DIPRecon method alternates between the LM-DRAMA algorithm and MR-DIP using an alternating direction method of multipliers. The evaluation shows that LM-DIPRecon achieves sharper images and better tradeoff curves between contrast and noise than other methods.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Computer Science, Interdisciplinary Applications
Siqi Li, Guobao Wang
Summary: This study proposes a deep kernel method for dynamic PET image reconstruction by utilizing a deep neural network to automatically learn an improved kernel model. Experimental results demonstrate that this method outperforms existing methods in terms of reconstruction performance.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Zahra Ashouri, Guobao Wang, Richard M. Dansereau, Robert A. DeKemp
Summary: Positron emission tomography (PET) imaging is used to track biochemical processes in the human body. This study evaluated a wavelet kernel method for PET image reconstruction and found that it performs better in contrast recovery and signal-to-noise ratio compared to the Gaussian kernel method.
IEEE TRANSACTIONS ON RADIATION AND PLASMA MEDICAL SCIENCES
(2022)
Article
Computer Science, Information Systems
Azra Tafro, Damir Sersic, Ana Sovic Krzic
Summary: This paper presents a novel method for PET image reconstruction by estimating Gaussian mixture model parameters. The method addresses the issues of computational complexity and noisy results in PET image reconstruction.
Article
Mathematics
Nicholas E. E. Protonotarios, George A. A. Kastis, Andreas D. D. Fotopoulos, Andreas G. G. Tzakos, Dimitrios Vlachos, Nikolaos Dikaios
Summary: This study focuses on improving motion correction algorithms in PET by using a motion-compensated image reconstruction (MCIR) algorithm based on a parabolic surrogate likelihood function. The parabolic surrogate algorithm converges faster than the expectation maximization (EM) algorithm, making it particularly useful in computationally demanding PET motion correction.
Article
Computer Science, Information Systems
Shuangliang Cao, Yuru He, Hongyan Zhang, Wenbing Lv, Lijun Lu, Wufan Chen
Summary: A novel PET image reconstruction framework was proposed in this study, aiming to enhance quantitative accuracy of individual frames using priors based on multiscale superpixel clusters. The method achieved substantial improvements in both visual and quantitative accuracy, outperforming other methods in studies on rats with PET scans.
Article
Biology
Shuangliang Cao, Yuru He, Hao Sun, Huiqin Wu, Wufan Chen, Lijun Lu
Summary: This study proposed a median nonlocal means (MNLM)-based kernel method for dynamic PET image reconstruction, which achieved visual and quantitative accuracy improvements. The MNLM kernel method outperformed other methods in both simulated low-count dynamic data and real low-dose F-18-FDG rat data in terms of noise reduction and intensity enhancement.
COMPUTERS IN BIOLOGY AND MEDICINE
(2021)
Article
Computer Science, Artificial Intelligence
Jiadong Zhang, Zhiming Cui, Caiwen Jiang, Shanshan Guo, Fei Gao, Dinggang Shen
Summary: This article proposes a learning-based method to reconstruct high-dose positron emission tomography (PET) images from low-dose PET images and corresponding total-body computed tomography (CT) images. The proposed hierarchical framework can consistently improve the performance of all body parts and outperforms the state-of-the-art methods in single-photon emission computed tomography (SPET) image reconstruction, with a peak signal-to-noise ratio (PSNR) of 30.6 dB for total-body PET images.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Interdisciplinary Applications
Kuang Gong, Ciprian Catana, Jinyi Qi, Quanzheng Li
Summary: This study proposes an unsupervised deep learning framework for direct parametric reconstruction from dynamic PET, which combines the PET statistical model, the patient's anatomical prior image, and the embedded linear kinetic model to achieve better reconstruction results.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Engineering, Electrical & Electronic
Pablo Galve, Jose Manuel Udias, Alejandro Lopez-Montes, Fernando Arias-Valcayo, Juan Jose Vaquero, Manuel Desco, Joaquin L. Herraiz
Summary: A novel method using super-iterations to exceed the resolution-noise limits in PET imaging was proposed. Improvement of approximately 10% in resolution and recovery coefficient was achieved while maintaining the same noise level. Qualitative results confirmed the enhancement in image quality from the proposed method.
IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING
(2021)
Article
Engineering, Biomedical
Fumio Hashimoto, Yuya Onishi, Kibo Ote, Hideaki Tashima, Taiga Yamaya
Summary: In this study, a DIP-based fully 3D PET image reconstruction method was proposed, which incorporated a forward-projection model into a loss function. The results showed that compared with other algorithms, the proposed method improved PET image quality and preserved the contrast of brain structures and inserted tumors.
PHYSICS IN MEDICINE AND BIOLOGY
(2023)
Article
Biochemical Research Methods
Palak Wadhwa, Kris Thielemans, Nikos Efthimiou, Kristen Wangerin, Nicholas Keat, Elise Emond, Timothy Deller, Ottavia Bertolli, Daniel Deidda, Gaspar Delso, Michel Tohme, Floris Jansen, Roger N. Gunn, William Hallett, Charalampos Tsoumpas
Summary: This research demonstrates successful computational and physical modelling of the PET-MR system for image acquisition, generating images comparable to those from the manufacturer; the new software developments will be integrated into STIR, providing researchers worldwide with opportunities to establish and expand their image reconstruction methods; by modelling all effects within the system matrix, PET images showing the metabolic uptake of administered radiopharmaceuticals were reconstructed accurately.
Article
Radiology, Nuclear Medicine & Medical Imaging
Guobao Wang, Lorenzo Nardo, Mamta Parikh, Yasser G. Abdelhafez, Elizabeth Li, Benjamin A. Spencer, Jinyi Qi, Terry Jones, Simon R. Cherry, Ramsey D. Badawi
Summary: Quantitative dynamic PET with compartmental modeling allows for multiparametric imaging and more accurate quantification. This study evaluates the necessity of voxelwise compartmental modeling strategies for total-body multiparametric imaging and finds that time delay correction and model selection improve the imaging results.
JOURNAL OF NUCLEAR MEDICINE
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Elizabeth J. Li, Benjamin A. Spencer, Jeffrey P. Schmall, Yasser Abdelhafez, Ramsey D. Badawi, Guobao Wang, Simon R. Cherry
Summary: This study investigated the use of pulse timing methods for estimating and correcting delay in total-body PET parametric imaging. The results showed that LE delay correction can be an efficient surrogate for JE, requiring a fraction of the computational time and allowing for rapid correction across more than 10^6 voxels in total-body PET datasets.
JOURNAL OF NUCLEAR MEDICINE
(2022)
Article
Computer Science, Interdisciplinary Applications
Tiantian Li, Mengxi Zhang, Wenyuan Qi, Evren Asma, Jinyi Qi
Summary: Respiratory motion is a significant factor affecting the quality of PET imaging. In this study, a robust joint estimation method combining deep learning with image registration was proposed. The method effectively estimated the emission image and patient motion from respiratory gated data, outperforming traditional registration-based methods. The proposed DL-ADMM algorithm showed promising results in both simulated and real data studies, reducing bias and improving image quality.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Interdisciplinary Applications
Kuang Gong, Ciprian Catana, Jinyi Qi, Quanzheng Li
Summary: This study proposes an unsupervised deep learning framework for direct parametric reconstruction from dynamic PET, which combines the PET statistical model, the patient's anatomical prior image, and the embedded linear kinetic model to achieve better reconstruction results.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Nimu Yuan, Shyam Rao, Quan Chen, Levent Sensoy, Jinyi Qi, Yi Rong
Summary: The study aimed to design a deep neural network architecture and loss function for improving soft-tissue contrast and preserving small anatomical features in ultra-low-dose cone-beam CTs of head and neck cancer imaging. The proposed network architecture and loss function effectively improved image quality in soft-tissue contrast, organ boundary, and small structure preservation for ultra-low-dose CBCT following Image Gently Protocol.
Article
Engineering, Biomedical
Edwin K. Leung, Yasser G. Abdelhafez, Eric Berg, Zhaoheng Xie, Xuezhu Zhang, Reimund Bayerlein, Benjamin Spencer, Elizabeth Li, Negar Omidvari, Aaron Selfridge, Simon R. Cherry, Jinyi Qi, Ramsey D. Badawi
Summary: This study assessed the relationship between image signal-to-noise ratio (SNR) and total-body noise-equivalent count rate (NECR) in a long axial field-of-view total-body PET system, and found that TOF-NECR may be a more accurate predictor of SNR for TOF-reconstructed images.
PHYSICS IN MEDICINE AND BIOLOGY
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Zahra Ashouri, Guobao Wang, Richard M. Dansereau, Robert A. DeKemp
Summary: Positron emission tomography (PET) imaging is used to track biochemical processes in the human body. This study evaluated a wavelet kernel method for PET image reconstruction and found that it performs better in contrast recovery and signal-to-noise ratio compared to the Gaussian kernel method.
IEEE TRANSACTIONS ON RADIATION AND PLASMA MEDICAL SCIENCES
(2022)
Article
Computer Science, Interdisciplinary Applications
Jinyi Qi, Bangyan Huang
Summary: Positron emission tomography (PET) is widely used in clinical and preclinical applications. This study proposes a new image reconstruction method to generate high-resolution positronium lifetime images using existing time-of-flight (TOF) PET scanners, overcoming the challenge of low spatial resolution. The proposed method allows for better understanding of tissue microenvironment and facilitates the study of disease mechanisms and treatment selection.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Interdisciplinary Applications
Siqi Li, Guobao Wang
Summary: This study proposes a deep kernel method for dynamic PET image reconstruction by utilizing a deep neural network to automatically learn an improved kernel model. Experimental results demonstrate that this method outperforms existing methods in terms of reconstruction performance.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Xinjie Zeng, Zipai Wang, Wanbin Tan, Eric Petersen, Xinjie Cao, Andy LaBella, Anthony Boccia, Dinko Franceschi, Mony de Leon, Gloria Chia-Yi Chiang, Jinyi Qi, Anat Biegon, Wei Zhao, Amir H. Goldan
Summary: This study presents a practical and cost-effective ultra-high resolution brain-dedicated PET scanner, which can accurately quantify radiotracer uptake in small regions by reducing the partial volume effect. The scanner performs well and has the potential to improve the diagnosis and treatment of neurological and oncological conditions with PET imaging.
Article
Computer Science, Interdisciplinary Applications
Siqi Li, Kuang Gong, Ramsey. D. D. Badawi, Edward J. Kim, Jinyi Qi, Guobao Wang
Summary: This paper proposes an implicit regularization method for kernel methods in image reconstruction of low-count PET data. By using a convolutional neural network to represent the kernel coefficient image in the PET forward model, the proposed method achieves better performance than existing methods. The neural KEM algorithm, which combines the KEM step and deep learning step, is demonstrated to be effective in both computer simulations and real patient data.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Chih-Chieh Liu, Yasser G. G. Abdelhafez, S. Paran Yap, Francesco Acquafredda, Silvia Schiro, Andrew L. Wong, Dani Sarohia, Cyrus Bateni, Morgan A. A. Darrow, Michele Guindani, Sonia Lee, Michelle Zhang, Ahmed W. W. Moawad, Quinn Kwan-Tai Ng, Layla Shere, Khaled M. M. Elsayes, Roberto Maroldi, Thomas M. M. Link, Lorenzo Nardo, Jinyi Qi
Summary: To address the challenges of data heterogeneity in tumor image segmentation, researchers proposed a deep learning-based Super Learner ensemble framework. By applying different data correction and normalization methods, the method improved the accuracy of tumor segmentation and mitigated performance instability and data heterogeneity.
JOURNAL OF DIGITAL IMAGING
(2023)
Proceedings Paper
Optics
Nimu Yuan, Jian Zhou, Jinyi Qi
Summary: A deep learning-based approach for enhancing the spatial resolution of sparse-view CT (SVCT) is proposed in this study. The proposed method utilizes a densely connected convolutional neural network (CNN) and a radial location map to recover the resolution loss and improve image quality.
MEDICAL IMAGING 2022: PHYSICS OF MEDICAL IMAGING
(2022)