Zhijun Chi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhijun Chi has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
Wenhui Huang
Tsinghua University
6 shared publications- Yingchao Du
Tsinghua University
6 shared publications Chuanxiang Tang
Tsinghua University
5 shared publications- Zhi Zhang
Key Laboratory of Particle and Radiation Imaging of Ministry of Education, Department of Engineering Physics, Tsinghua University, Beijing, China
5 shared publications Chuan-Xiang Tang
Key Laboratory of Particle and Radiation Imaging of Ministry of Education, Department of Engineering Physics, Tsinghua University, Beijing, China
4 shared publications- Chuanxiang Tang
Tsinghua University
4 shared publications
Latest Publications
- 30 October 2026
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (21) , 215003
Physics-informed neural network for photoacoustic tomography reconstruction
Objective. Conventional photoacoustic tomography (PAT) reconstruction algorithms often suffer from performance degradation under complex media and non-ideal detection conditions. To tackle these limitations, we propose a novel reconstruction method based on physics-informed neural networks (PINNs). Approach. The network model is optimized with constraints from observational data, wave equation, and initial conditions, to reconstruct the spatial distribution of initial acoustic pressure. The PINN framework further incorporates time reversal method to improve computational efficiency. Main results. Comprehensive PAT simulations demonstrate that the PINN achieves superior reconstruction quality compared with conventional algorithms. The PINN exhibits favorable adaptivity to both heterogeneous and lossy media, and strong robustness against noisy measurements, sparse sensor layouts and limited-view detection geometries. Significance. The PINN integrates the physical interpretability and the powerful capability of neural networks to represent continuous acoustic pressure fields. It eliminates the requirement for training datasets, offering broad application potential for PAT.
- 30 October 2026
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 71 (21) , 215003
Physics-informed neural network for photoacoustic tomography reconstruction
Objective. Conventional photoacoustic tomography (PAT) reconstruction algorithms often suffer from performance degradation under complex media and non-ideal detection conditions. To tackle these limitations, we propose a novel reconstruction method based on physics-informed neural networks (PINNs). Approach. The network model is optimized with constraints from observational data, wave equation, and initial conditions, to reconstruct the spatial distribution of initial acoustic pressure. The PINN framework further incorporates time reversal method to improve computational efficiency. Main results. Comprehensive PAT simulations demonstrate that the PINN achieves superior reconstruction quality compared with conventional algorithms. The PINN exhibits favorable adaptivity to both heterogeneous and lossy media, and strong robustness against noisy measurements, sparse sensor layouts and limited-view detection geometries. Significance. The PINN integrates the physical interpretability and the powerful capability of neural networks to represent continuous acoustic pressure fields. It eliminates the requirement for training datasets, offering broad application potential for PAT.
- 7 October 2026
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 37 (11) , 238
Triple-energy computed tomography in the megavoltage range with an inverse Compton scattering light source
- 7 October 2026
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 37 (11) , 238
Triple-energy computed tomography in the megavoltage range with an inverse Compton scattering light source
- 28 July 2026
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 37 (10) , 208
Laser system of the very compact inverse Compton scattering γ-ray source at Tsinghua University
- 28 July 2026
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 37 (10) , 208
Laser system of the very compact inverse Compton scattering γ-ray source at Tsinghua University
- 1 June 2026
- Published by AIP Publishing in Review of Scientific Instruments
- Vol. 97 (6)
A novel method for measuring the energy spectrum of an inverse Compton scattering source based on nuclear resonance fluorescence
Inverse Compton scattering (ICS) gamma ray sources are capable of producing quasi-monochromatic, continuously tunable, high-brightness, precisely polarization-controllable, and ultrashort gamma ray pulses in the energy range from tens of keV to several MeV or even higher. The energy spectrum measurement of an ICS source not only serves as a key indicator of the device operating status but also provides foundational information. In this paper, we proposed a novel method of using nuclear resonance fluorescence (NRF) as a probe for spectrum measurements. By utilizing the continuous tunability of an ICS source, NRF photons can be excited at different points across the spectrum. The shape of the energy spectrum can then be effectively scanned and reconstructed by recording the relative NRF yields at different energy points. The feasibility of the proposed method was validated by Geant4 simulations of measuring NRF photon emission from 56Fe irradiated by an ICS source. We utilized the 845 and 3449 keV NRF photons of 56Fe to measure a Gaussian spectrum and a segmented Gaussian spectrum, which correspond to ICS spectra under different collection angles. The simulation results showed high precision for quasi-monochromatic gamma ray spectrum measurements, with a normalized root mean square error of less than 5%. To maintain a sufficient signal-to-noise ratio during the measurement, the energy resolution of detectors is suggested to be less than 1% of the energy being measured. Given an energy tuning precision of ΔE, the minimum measurable width of the energy spectrum, in terms of standard deviation, can reach 0.85 ΔE. Further experiments are needed to validate its practical feasibility.
- 1 June 2026
- Published by AIP Publishing in Review of Scientific Instruments
- Vol. 97 (6)
A novel method for measuring the energy spectrum of an inverse Compton scattering source based on nuclear resonance fluorescence
Inverse Compton scattering (ICS) gamma ray sources are capable of producing quasi-monochromatic, continuously tunable, high-brightness, precisely polarization-controllable, and ultrashort gamma ray pulses in the energy range from tens of keV to several MeV or even higher. The energy spectrum measurement of an ICS source not only serves as a key indicator of the device operating status but also provides foundational information. In this paper, we proposed a novel method of using nuclear resonance fluorescence (NRF) as a probe for spectrum measurements. By utilizing the continuous tunability of an ICS source, NRF photons can be excited at different points across the spectrum. The shape of the energy spectrum can then be effectively scanned and reconstructed by recording the relative NRF yields at different energy points. The feasibility of the proposed method was validated by Geant4 simulations of measuring NRF photon emission from 56Fe irradiated by an ICS source. We utilized the 845 and 3449 keV NRF photons of 56Fe to measure a Gaussian spectrum and a segmented Gaussian spectrum, which correspond to ICS spectra under different collection angles. The simulation results showed high precision for quasi-monochromatic gamma ray spectrum measurements, with a normalized root mean square error of less than 5%. To maintain a sufficient signal-to-noise ratio during the measurement, the energy resolution of detectors is suggested to be less than 1% of the energy being measured. Given an energy tuning precision of ΔE, the minimum measurable width of the energy spectrum, in terms of standard deviation, can reach 0.85 ΔE. Further experiments are needed to validate its practical feasibility.
- 15 May 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (3)
Accurate Monte Carlo simulation and image reconstruction for Compton cameras
Objective. The expanding applications of Compton cameras in biomedical imaging demand precise simulation and reconstruction methods that incorporate realistic physics to optimize performance. To address this need, this study develops an integrated computational framework that combines a Geant4-based Monte Carlo simulation with a dedicated image reconstruction algorithm. Approach. Compared with previous approaches, this framework accurately models inter-pixel energy deposition crosstalk and depth of interaction in Monte Carlo simulation, while also accounting for multiple Compton scattering. In addition, it incorporates accurate models into the system matrix calculation within the list-mode maximum-likelihood expectation-maximization image reconstruction algorithm, including the Compton scattering cross section, the spatial resolution of both the scatterer and absorber detectors, and Doppler broadening effects. Main Results. The framework was validated through simulations of a proposed medical Compton camera design employing a silicon scatterer and a cadmium zinc telluride absorber. Compared with conventional simplified models, the proposed accurate system matrix model yields a marked improvement in reconstruction quality across diverse source geometries, including ideal point, planar, and line sources, as well as hot sources with a uniform background. Significance. This work establishes a high-fidelity framework that enables both physically realistic simulation and quantitatively accurate image reconstruction for Compton cameras. By rigorously validating the model's advantage across multiple source types, it provides an essential tool for advancing the design, optimization, and practical application of Compton imaging systems in biomedical imaging.
- 15 May 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (3)
Accurate Monte Carlo simulation and image reconstruction for Compton cameras
Objective. The expanding applications of Compton cameras in biomedical imaging demand precise simulation and reconstruction methods that incorporate realistic physics to optimize performance. To address this need, this study develops an integrated computational framework that combines a Geant4-based Monte Carlo simulation with a dedicated image reconstruction algorithm. Approach. Compared with previous approaches, this framework accurately models inter-pixel energy deposition crosstalk and depth of interaction in Monte Carlo simulation, while also accounting for multiple Compton scattering. In addition, it incorporates accurate models into the system matrix calculation within the list-mode maximum-likelihood expectation-maximization image reconstruction algorithm, including the Compton scattering cross section, the spatial resolution of both the scatterer and absorber detectors, and Doppler broadening effects. Main Results. The framework was validated through simulations of a proposed medical Compton camera design employing a silicon scatterer and a cadmium zinc telluride absorber. Compared with conventional simplified models, the proposed accurate system matrix model yields a marked improvement in reconstruction quality across diverse source geometries, including ideal point, planar, and line sources, as well as hot sources with a uniform background. Significance. This work establishes a high-fidelity framework that enables both physically realistic simulation and quantitatively accurate image reconstruction for Compton cameras. By rigorously validating the model's advantage across multiple source types, it provides an essential tool for advancing the design, optimization, and practical application of Compton imaging systems in biomedical imaging.
- 24 March 2026
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 139 (12)
Imaging efficiency optimization for x-ray fluorescence computed tomography based on Thomson scattering x-ray sources
X-ray fluorescence computed tomography (XFCT) based on Thomson scattering x-ray sources is a novel molecular imaging technique applicable in both laboratory and clinical settings. However, its practical development is limited by relatively low imaging efficiency. To address this challenge, this study presents a dual approach that simultaneously optimizes the incident x-ray energy to maximize fluorescence yield and designs a three-pinhole collimator to enhance detection sensitivity. For the element gold (Au) widely used in XFCT imaging, the optimal incident x-ray energy was determined to be 82.6 keV for a Thomson scattering x-ray source with a root mean square bandwidth of 1%. Using contrast agents containing gold solutions with Au concentrations ranging from 0.1 to 4.0 wt. %, the performance of the designed imaging system was evaluated by Monte Carlo simulations. Compared to the conventional single-pinhole collimator, the three-pinhole design provides an approximately 2.5-fold enhancement in sensitivity and lowers the detection limit for gold solution from 0.183 to 0.161 wt. %. By increasing the incident x-ray energy to 100 keV, the scatter background was significantly reduced; however, lower contrast-to-noise ratios were observed across all gold solution concentrations, owing to the reduction of fluorescence yield at elevated x-ray energies. For the designed imaging system, a spatial resolution of approximately 0.7 mm can be achieved. Using this system, a complete XFCT scan can be completed in approximately 24 s using a Thomson scattering x-ray source with a photon yield of 1013 photons/s, without inducing pulse pileup effects; meanwhile, gold solutions with a concentration as low as 0.2 wt. % can be clearly resolved at a radiation dose of 2 Gy.
- 24 March 2026
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 139 (12)
Imaging efficiency optimization for x-ray fluorescence computed tomography based on Thomson scattering x-ray sources
X-ray fluorescence computed tomography (XFCT) based on Thomson scattering x-ray sources is a novel molecular imaging technique applicable in both laboratory and clinical settings. However, its practical development is limited by relatively low imaging efficiency. To address this challenge, this study presents a dual approach that simultaneously optimizes the incident x-ray energy to maximize fluorescence yield and designs a three-pinhole collimator to enhance detection sensitivity. For the element gold (Au) widely used in XFCT imaging, the optimal incident x-ray energy was determined to be 82.6 keV for a Thomson scattering x-ray source with a root mean square bandwidth of 1%. Using contrast agents containing gold solutions with Au concentrations ranging from 0.1 to 4.0 wt. %, the performance of the designed imaging system was evaluated by Monte Carlo simulations. Compared to the conventional single-pinhole collimator, the three-pinhole design provides an approximately 2.5-fold enhancement in sensitivity and lowers the detection limit for gold solution from 0.183 to 0.161 wt. %. By increasing the incident x-ray energy to 100 keV, the scatter background was significantly reduced; however, lower contrast-to-noise ratios were observed across all gold solution concentrations, owing to the reduction of fluorescence yield at elevated x-ray energies. For the designed imaging system, a spatial resolution of approximately 0.7 mm can be achieved. Using this system, a complete XFCT scan can be completed in approximately 24 s using a Thomson scattering x-ray source with a photon yield of 1013 photons/s, without inducing pulse pileup effects; meanwhile, gold solutions with a concentration as low as 0.2 wt. % can be clearly resolved at a radiation dose of 2 Gy.
- 12 March 2026
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 37 (5) , 92
An effective method toward large field-of-view gamma-ray computed tomography based on an inverse Compton scattering light source
- 12 March 2026
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 37 (5) , 92
An effective method toward large field-of-view gamma-ray computed tomography based on an inverse Compton scattering light source
- Open Access
- 23 September 2025
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 138 (12)
Simulation of x-ray fluorescence ghost imaging by inverse Compton scattering source: A feasibility study
X-ray fluorescence computed tomography (XFCT) has shown great potential in molecular biomedical imaging. However, conventional XFCT has some limitations, including sensitivity and a trade-off between spatial resolution and dose. X-ray fluorescence ghost imaging (XRF-GI) makes it possible to realize high-resolution imaging with low doses. However, the efficient implantation of XRF-GI requires a quasi-monochromatic x-ray source. The inverse Compton scattering (ICS) source could meet the requirement well. In this study, we developed a Monte Carlo simulation model of an XRF-GI system based on the Geant4 toolkit. The simulation model was composed of an ICS source, a series of Hadamard pattern-based masks as a modulating system, a single-pixel energy-resolving detector, and a phantom embedded with gold nanoparticles (GNPs) of different concentrations as the contrast agent. The image reconstructed by the total variation-regularized least squares (L2-TV) algorithm was evaluated by four criteria, including the mean square error, the contrast-to-noise ratio, the coefficient of determination (R2), and the limit of detection. Compared to XFCT, the reconstructed image of XRF-GI has better performance in all four criteria, and the dose of XRF-GI was halved without a decrease in resolution. It was validated that the beam properties of the ICS source fitted well with the implantation of XRF-GI for different conditions. In conclusion, the XRF-GI by an ICS source has great potential for biomedical imaging.
- 23 September 2025
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 138 (12)
Open AccessSimulation of x-ray fluorescence ghost imaging by inverse Compton scattering source: A feasibility study
X-ray fluorescence computed tomography (XFCT) has shown great potential in molecular biomedical imaging. However, conventional XFCT has some limitations, including sensitivity and a trade-off between spatial resolution and dose. X-ray fluorescence ghost imaging (XRF-GI) makes it possible to realize high-resolution imaging with low doses. However, the efficient implantation of XRF-GI requires a quasi-monochromatic x-ray source. The inverse Compton scattering (ICS) source could meet the requirement well. In this study, we developed a Monte Carlo simulation model of an XRF-GI system based on the Geant4 toolkit. The simulation model was composed of an ICS source, a series of Hadamard pattern-based masks as a modulating system, a single-pixel energy-resolving detector, and a phantom embedded with gold nanoparticles (GNPs) of different concentrations as the contrast agent. The image reconstructed by the total variation-regularized least squares (L2-TV) algorithm was evaluated by four criteria, including the mean square error, the contrast-to-noise ratio, the coefficient of determination (R2), and the limit of detection. Compared to XFCT, the reconstructed image of XRF-GI has better performance in all four criteria, and the dose of XRF-GI was halved without a decrease in resolution. It was validated that the beam properties of the ICS source fitted well with the implantation of XRF-GI for different conditions. In conclusion, the XRF-GI by an ICS source has great potential for biomedical imaging.
- Open Access
- 15 October 2024
- Published by American Physical Society (APS) in Physical Review Accelerators and Beams
- Vol. 27 (10)
Monte Carlo simulation code for polarized inverse Compton scattering sources
Polarized x rays or gamma rays generated from inverse Compton scattering (ICS) sources are high-brightness, quasimonochromatic, and polarization-tunable and are widely applied in medical imaging, material science, and nuclear physics research. In this paper, to simulate ICS between electron beams and polarized laser beams, a Monte Carlo simulation code has been developed and implemented based on the 4 toolkit. This code considers the polarization as well as the bandwidth of incident laser beams. The code is benchmarked by comparing its simulation results with those from the well-known Monte Carlo tool called and experimental results. The comparisons show that this code accurately reproduces the scattered photons, demonstrating its potential for various ICS-based applications. Published by the American Physical Society 2024
- 15 October 2024
- Published by American Physical Society (APS) in Physical Review Accelerators and Beams
- Vol. 27 (10)
Open AccessMonte Carlo simulation code for polarized inverse Compton scattering sources
Polarized x rays or gamma rays generated from inverse Compton scattering (ICS) sources are high-brightness, quasimonochromatic, and polarization-tunable and are widely applied in medical imaging, material science, and nuclear physics research. In this paper, to simulate ICS between electron beams and polarized laser beams, a Monte Carlo simulation code has been developed and implemented based on the 4 toolkit. This code considers the polarization as well as the bandwidth of incident laser beams. The code is benchmarked by comparing its simulation results with those from the well-known Monte Carlo tool called and experimental results. The comparisons show that this code accurately reproduces the scattered photons, demonstrating its potential for various ICS-based applications. Published by the American Physical Society 2024
- 21 September 2024
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 35 (10) , 1-10
Simultaneous fluorescence and Compton scattering computed tomography based on linear polarization X-ray
- 21 September 2024
- Published by Springer Nature in Nuclear Science and Techniques
- Vol. 35 (10) , 1-10
Simultaneous fluorescence and Compton scattering computed tomography based on linear polarization X-ray