Development of an Omnidirectional Compton Camera with Enhanced Energy Resolution for Radioactive Source Localization

Xin-Yu Yang¹, Jian Yang¹,²,*, Guo-Qiang Zeng¹,², Hao-Wen Deng¹, Cheng-Shuai Tian¹, Chuan-Hao Hu¹,²

¹ College of Nuclear Technology and Automation Engineering, Chengdu University of Technology, Chengdu 610059, China
² Key Laboratory of Applied Nuclear Techniques in Geosciences Sichuan, Chengdu University of Technology, Chengdu 610059, China

Corresponding author. E-mail address: yang_jian@cdut.edu.cn

Abstract

For radioactive source localization, traditional Compton cameras are limited by narrow imaging fields of view and low detection efficiency, resulting in slow response times in complex environments. To address the limitations in azimuthal sensitivity of conventional dual-layer scintillator detectors and the restricted sensitive volume of semiconductor detectors, this study proposes an omnidirectional Compton camera based on four-layer scintillator detectors arranged in a square ring-shaped structure. This imaging system achieves omnidirectional sensitivity with continuous 360° azimuthal response, significantly enhancing detection efficiency by minimizing the escape of scattered photons. The smallest imaging unit utilizes an independent CsI(Tl) crystal coupled with a single SiPM, which reduces light propagation loss and improves light collection efficiency to enhance energy resolution. To simplify the electronic systems, a serial readout circuit is employed to reduce the number of readout channels. Experimental results demonstrate that the total energy resolution for the ¹³⁷Cs source was 6.2% (FWHM) after position and energy calibration. Using the maximum likelihood expectation maximization algorithm for image reconstruction, the imaging system achieves accurate localization of radioactive sources with an angular resolution of 12° and can clearly distinguish the positions of two radioactive sources.

Keywords: Compton imaging; Omnidirectional sensitivity; Scintillator detector; Serial readout circuit; Energy resolution

1. Introduction

With the widespread application of radioactive materials, radioactive source localization and imaging techniques have become crucial for nuclear decommissioning, medical imaging, and environmental monitoring [1-7]. Current gamma-ray imaging techniques, such as coded aperture imaging and Compton imaging, exhibit limitations in directional sensitivity and angular response that make them difficult to adapt to complex environments. Especially in scenarios such as nuclear facilities, where the types, energies, and spatial distribution of radioactive materials are unknown, existing imaging techniques still face challenges in terms of directional sensitivity and detection efficiency. Although monolithic semiconductor detectors offer a full field of view and high energy resolution, they are constrained by their small volume and high cost, making it difficult to image weak radioactive sources over large fields. For example, the H100 Compton camera developed by H3D Inc. supports 4π field-of-view imaging with an angular resolution of 20°, but its sensitive volume is only 6 cm³ [10].

Scintillator detectors demonstrate significant application prospects in gamma-ray imaging due to their advantages of large sensitive volume, low cost, and high detection efficiency [11,12]. These detectors typically employ a dual-layer pixelated structure or a design combining a coded aperture mask with a single-layer scintillator. For example, Waseda University proposed a Compton camera that employs a pixelated GAGG scintillator coupled with an MPPC array, achieving an energy resolution of 7.8% for 662 keV gamma rays and an angular resolution of about 8° [13]. To further enhance detection efficiency and imaging quality, researchers have developed various structural optimizations based on the traditional scatterer-absorber dual-layer structure, such as adopting monolithic scintillator designs, enlarging the absorber volume, and increasing the number of scatterer layers [14,15], to achieve omnidirectional detection capability and higher detection efficiency. In 2022, H. Lee et al. developed a Compton camera using a monolithic GAGG scintillator coupled with dual-sided SiPM readout, which eliminated the need to distinguish between scatterer and absorber layers. This design achieved a 4π imaging field of view and effectively reduced radioactive background noise. However, limitations imposed by the relatively small detector volume, combined with relatively poor energy resolution, compromise its practical applicability to some extent. In 2023, researchers from Beijing Normal University proposed a modified Compton camera structure that adopted an annular absorber layer instead of the traditional planar layout, significantly improving detection efficiency for distant radioactive sources [16]. In 2022, Xiuzuo Liang from the Institute of High Energy Physics designed a spherical detector system based on icosahedral symmetry [17]. This system achieved 4π omnidirectional imaging by uniformly distributing 80 GAGG scintillator pixels across its spherical surface and integrated both coded aperture and Compton imaging modes, ensuring wide-energy imaging capability. However, both the annular absorber layer and the spherical detector pose limitations for practical real-world application. Despite significant advances in recent research, there remains room for improvement in azimuthal sensitivity and detection efficiency.

In the development of scintillator imaging detectors, achieving an optimal balance between crystal size, number of readout channels, and energy resolution is a key challenge. To reduce the complexity of readout electronics, researchers have proposed various shared readout schemes that decrease the number of readout channels and system costs while maintaining imaging performance [18]. Among these, Discretized Positioning Circuit (DPC) achieves row-column signal diversion through a resistor network and uses differential gain design across four readout channels for position encoding, reducing the readout channel number from N² to a fixed four channels. This approach significantly lowers hardware costs and simplifies system complexity. However, as the detector array size increases, DPC faces critical challenges such as geometric distortion and degradation of the signal-to-noise ratio. In 2021, Zhu Balin replaced the traditional DPC with a row/column readout circuit, reducing the readout channel number from N² to 2N [19]. This modification effectively reduces the impact of multiple Compton scattering events on imaging quality, significantly improving imaging performance. Although the Symmetric Charge Division Circuit can simplify the output of the row/column readout circuit to four signal outputs, it is unable to effectively eliminate multiple Compton scattering events occurring in the scatterer layer, leading to a decline in reconstructed image quality.

This study proposes an omnidirectional Compton camera based on a square ring-shaped structure, aiming to enhance the detection performance and spatial sensitivity of the imaging system. The system establishes omnidirectional sensitivity with continuous 360° azimuthal response through the square ring-shaped structure, significantly enhancing detection efficiency by minimizing the escape of scattered photons. Additionally, the system employs a serial readout circuit, which reduces the number of readout channels while maintaining excellent energy resolution.

2.1 Structure of Imaging Detector

To improve the detection efficiency and spatial sensitivity of the Compton camera, we designed an omnidirectional Compton imaging detector. The detector employs a square ring-shaped structure, as shown in Figure 1 FIGURE:1, composed of four independent planar detectors arranged on the front, back, left, and right sides. This design significantly expands the effective field of view of the imaging detector, achieving omnidirectional sensitivity with 360° azimuthal angular coverage. Moreover, the four-layer separated structure enhances the absorption efficiency of scattered photons by reducing photon escape probability from the sensitive volume, thereby improving overall detection efficiency.

When selecting a scintillator detector, it is essential to consider both the energy resolution of the detector and the limitations of current fabrication processes. CsI(Tl) crystals are widely regarded as ideal scintillator materials due to their high light yield and excellent energy resolution, which makes them highly effective for various radiation detection applications. In addition to these key technical advantages, they exhibit outstanding physicochemical robustness, particularly in terms of moisture resistance and environmental adaptability. This robustness enables stable long-term operation in complex environments such as nuclear facility monitoring. Furthermore, the cost-effectiveness resulting from their mature fabrication process makes CsI(Tl) an optimal choice that offers a favorable balance between performance and practicality. To achieve position resolution for CsI(Tl) crystals, a modular detector array is typically used to construct a two-dimensional (2-D) position-sensitive planar detector, which plays a critical role in accurately localizing gamma-ray interactions.

Theoretically, increasing the number of detectors can significantly enhance spatial resolution, but it also substantially increases the number of readout channels, which in turn raises the complexity and cost of the imaging system. Therefore, under the premise of ensuring imaging sensitivity, it is crucial to optimize the number and layout of detectors rationally. Considering these factors, this study adopts an 8 × 8 pixel array for the planar detector, resulting in a total of 256 detectors for the omnidirectional Compton camera, a number deemed acceptable for practical implementation.

The operational principle of the omnidirectional Compton imaging detector is shown in Figure 1(b). It employs the principle of Compton scattering to achieve imaging of radioactive sources. Compared to conventional dual-layer detectors, the four-layer detectors in the omnidirectional Compton camera are all capable of detecting the spatial coordinates of both Compton scattering and photoelectric absorption points. By combining the spatial coordinates of scattering and absorption points, the direction of the scattered photon can be reconstructed, and the Compton scattering angle can be determined from the deposited energy. This method enables efficient localization of radioactive sources without the need for mechanical collimators. Additionally, the four-layer detector design significantly improves the sensitivity of the Compton camera in all directions and substantially enhances the collection efficiency of scattered photons by minimizing their escape.

In the omnidirectional Compton camera, the four-layer detectors with 2-D position sensitivity record the interaction positions ((x₁, y₁, z₁) and (x₂, y₂, z₂)) and deposited energies (E₁ and E₂) of both the incident photon and the scattered photon within the detector. If the incident photon scatters in one detector layer and deposits its full energy in another layer, the Compton scatter angle θₛ can be calculated using the energy loss of the photon before and after scattering, as follows:

$$
\langle MATH_0 \rangle
$$

where E₀ is the energy of the incident photon, E₁ and E₂ are the deposited energies of the incident photon and scattered photon in the detector, mₑ is the electron mass, and c is the speed of light. According to the deposited energy and scattering angle, as shown in Figure 1(a), a Compton cone can be constructed. This cone has the extension line of the two interaction positions as its axis, with the scattering point as the apex and the scattering angle as the cone angle. As shown in Figure 1(b), after recording a large number of Compton scattering events, localization of the radioactive source can be determined by superimposing multiple Compton cones obtained from different scattering events.

2.2 Signal Readout and Sampling Circuits

The block diagram of the circuit system is shown in Figure 2 FIGURE:2. This system employs a cascaded signal processing architecture, enabling conversion from analog to digital signals. To reduce the number of readout channels in the imaging system, four serial readout circuits are used to acquire signals from a planar detector. These exponentially decaying signals are then sampled and processed by an 8-channel digital acquisition system. Finally, the data from the four-layer detectors is aggregated through a 32-channel data transfer board and transmitted to the host computer via an Ethernet interface for reception and processing.

As shown in Figure 2(b), each detector consists of a 6.0 mm × 6.0 mm × 6.0 mm CsI(Tl) crystal coupled with a SiPM (EQR2011-6060D-S). The independent coupling design, where the SiPM is directly coupled to the CsI(Tl) crystal, enhances photon collection efficiency while suppressing electronic noise, thereby ensuring high energy resolution. To further improve light output efficiency, all surfaces of the crystal except the bottom light output surface are covered with a 0.3 mm thick TiO₂ reflective layer. This layer uses total internal reflection to guide scintillation photons toward the light output surface, significantly improving light output efficiency and reducing optical crosstalk between adjacent pixels. The packaged CsI(Tl) detector has dimensions of 6.6 mm × 6.6 mm × 6.0 mm. The bottom surface serves as the light output surface, which is optically coupled to the SiPM using silicone oil. The refractive index matching property of the silicone oil minimizes photon loss at the interface, thereby improving scintillation photon conversion efficiency.

Figure 2(c) shows the structure of the serial readout circuit. It consists of n (n > 2) charge-sensitive preamplifiers and n-1 serial resistors (Rₛ) forming the SiPM signal readout network. Each SiPM signal is DC-coupled into the serial resistor network to ensure effective signal extraction from every readout channel. This innovative design allows multiple independent SiPM signals to be read through only two readout channels, significantly reducing the complexity of the backend signal processing system. To balance the requirements of the imaging system for energy resolution, position resolution, and array scalability, a 4 × 4 SiPM array module was designed based on the serial readout circuit [20]. By assembling four SiPM array modules, an equivalent 8 × 8 planar detector system can be constructed, requiring only 8 readout channels. This design drastically reduces the number of readout channels compared to the conventional requirement of 64 readout channels, enhancing the flexibility and practicality of system expansion.

To achieve processing of planar detector readout signals, this paper designs an 8-channel digital acquisition system that includes gain adjustment and DC offset adjustment, the structure of which is shown in Figure 2(d). The pulse signals from the serial readout circuit are first processed by a signal conditioning circuit to optimize the dynamic range and maximize utilization of the ADC's sampling range. The signals are then further conditioned by a differential amplifier circuit, which converts the single-ended inputs to differential outputs to suppress common-mode noise and drive the ADC. The conditioned signals are sampled by a high-precision ADC (8-channel, 12-bit, 80 Msps) to achieve analog-to-digital conversion. The digitized data is transmitted via an LVDS interface to the FPGA, where trigger logic control, timestamping, and data packaging are implemented. Finally, the data from the four-layer detectors is transmitted in parallel to a 32-channel data transfer board, which aggregates the data from various digital acquisition modules and provides both analog and digital power. The aggregated data is subsequently transmitted to the host computer via Ethernet.

3.1 Digital Signal Processing Method

In the serial readout circuit, the serial resistor network causes charge partitioning, resulting in distinct pulse signals being generated at both ends of the readout circuit. Therefore, verifying the temporal coincidence of the two signals within a defined coincidence window is essential to confirm their origin from the same particle interaction event. As shown in Figure 3 FIGURE:3, when gamma rays strike the detector, pulse signals are simultaneously generated at both ends of the serial readout circuit. To prevent the loss of weak signals, the FPGA employs a forced triggering method to capture the signal waveform. This method ensures that whenever one signal is triggered, the other is captured synchronously, thereby preserving data integrity.

To accurately extract amplitude and time information from nuclear pulse signals, this system implements a dual-channel signal processing architecture comprising a fast shaping channel and a slow shaping channel. The fast shaping channel is primarily used for pulse triggering and pile-up identification. It employs a symmetric zero-area trapezoidal shaping algorithm, which not only enables automatic baseline restoration but also suppresses baseline fluctuations through its symmetric shape [21]. The slow shaping channel typically employs trapezoidal shaping for pulse amplitude extraction. This algorithm effectively enhances the signal-to-noise ratio, thereby ensuring accurate acquisition of amplitude information. The complete recursive formula of trapezoidal shaping is shown below:

$$
\langle MATH_1 \rangle
$$

where Tₛ is the sampling period and τ is the RC time constant. Figure 3(b) shows the waveform after trapezoidal shaping. The shaped waveform exhibits a distinct trapezoidal shape, with stable plateaus at the upper and lower levels. By calculating the difference between the average values of the upper and lower plateaus, the amplitude value Aᵢ, which is used to calculate particle position and energy, is obtained.

The hardware system employs a packet-based data processing method to enable comprehensive particle data collection. In this mode, each valid pulse generates a unique particle data packet containing information such as the channel number where the particle interacted with the detector, the time of interaction, and the signal amplitude. The particle mode ensures complete particle data collection, which is then packaged and transmitted to a computer for further processing. In addition, the particle mode allows real-time callbacks to the database for reprocessing the data to obtain new results, thus avoiding repeated measurements.

3.2 Position and Energy Reconstruction

As shown in Figure 4 FIGURE:4, an omnidirectional Compton imaging experimental system was developed in this study, consisting of a scintillator detector array [Figure 4(b)] and signal readout circuit [Figure 4(c)]. The gamma radioactive source was placed approximately 1 meter away from the center of the detector to ensure far-field imaging geometry, which facilitates the accuracy and reliability of subsequent image reconstruction.

In the serial readout circuit, multiple detectors share the serial resistor network, making it impossible to directly determine the interaction position of the incident particle. Therefore, the amplitudes A₁ and A₂ of the output signals at both ends of the circuit are used to infer the interaction position of the particle within the detector array. This method is based on the physical principle that signal amplitude decreases as the distance between the particle interaction point and the readout end increases. By comparing the relative amplitudes of the signals at both ends, the interaction position of the particle within the detector array can be inferred. During the experiment, after acquiring the particle data packets, coincidence matching is performed using channel numbers and timestamps to identify two correlated events triggered by the same particle. Figure 5 FIGURE:5 shows the distribution of output signal amplitudes from the four serial readout circuits in the planar detector. The scatter diagrams show 64 linear distributions in the form of bars, with each distribution representing the total amplitude measured by an individual scintillator crystal in a planar detector. These diagrams provide a visual representation of the relationship between signal amplitude and particle interaction position.

To evaluate the position resolution of the detector array using the serial readout circuit, the F value for each event is calculated according to Equation 7, and statistical analysis is performed on the F value for different positions. Figure 5(c) and Figure 5(d) show the distribution diagrams of amplitude ratio F for events with energy above 250 keV and for full-energy peak events. It can be observed from the figures that the F value clearly differentiates between responses from different detectors, thereby validating that the serial readout circuit has position resolution capability. A comparison between the distributions in Figure 5(c) and Figure 5(d) shows that as energy increases, the intervals of the amplitude ratio become clearer. Consequently, the F value distribution derived from full-energy peak events exhibits higher position resolution capability. Based on these results, a mapping table between the amplitude ratio F and detector position was constructed using full-energy peak events. Using this table, the position of each incident particle can be rapidly determined from the measured F value after each event. Therefore, the position resolution of the system is primarily limited by the detector size.

Due to gain differences among components such as the detector crystal, SiPM, and readout electronics, the position of the full-energy peak shifts significantly. Therefore, spectrum shift correction is required to achieve optimal overall energy spectrum performance. The linear correction formula is shown below:

$$
\langle MATH_2 \rangle
$$

where Eᵧ is the theoretical target energy, Aₚᵢ is the energy corresponding to the full-energy peak in the energy spectrum of the i-th detector, and kᵢ and bᵢ are the calibration factors for the corresponding channel. Figure 5(b) shows the scatter diagrams after energy calibration. The full-energy peak positions of all detectors are aligned, which facilitates synthesis of the total energy spectrum and enhances the energy resolution and stability of the overall system. After position and energy calibration, the interaction positions and deposited energies of the incident particle and scattered photons within the detectors are obtained, which serve as the basis for subsequent imaging event selection.

3.3 Imaging Event Selection and Imaging Algorithm

The available Compton imaging event involves a single Compton scattering event, where the gamma-ray photon first scatters in the scattering detector and is subsequently photoelectrically absorbed in the absorption detector. Therefore, to achieve Compton imaging, we need to select valid imaging events. The timestamps in the particle data packet record the precise moment of interaction between the particle and the detector and serve as the basis for identifying coincident events across multiple channels. After chronological sorting of all channel data, the system identifies coincident events in other channels that occur within a predefined time window for each event in the primary channel. If a coincidence is identified, the event is classified as valid.

In the omnidirectional Compton camera, a single Compton scattering event is characterized by the scattering and absorption processes occurring in different SiPM array modules within the four-layer detectors. To ensure that these two physical processes belong to the same scattering event, a coincidence time window is established. Based on analysis of the timing differences between signals triggered in different SiPM array modules, the coincidence time window is set to 0.5 μs. Meanwhile, the energy peak is used to identify the radionuclide species in the radioactive source, and only events within the energy range near the full-energy peak are selected for imaging. By combining coincidence event selection with energy windowing, Compton scattering events can be effectively selected, thereby enhancing imaging accuracy and efficiency.

After acquiring sufficient Compton imaging events, it becomes critical to select an appropriate image reconstruction algorithm to ensure high-accuracy localization. To meet the high-precision requirements for radioactive source localization in complex nuclear facility environments, we selected the maximum likelihood expectation maximization (MLEM) algorithm as the reconstruction method after comprehensive assessment of factors such as reconstruction accuracy, computational efficiency, and system implementation complexity. Although the simple back-projection algorithm offers fast real-time imaging capabilities, its angular resolution is relatively low. In contrast, the MLEM algorithm significantly improves angular resolution through iterative processing of Compton imaging events, allowing for more accurate radioactive source localization. The MLEM algorithm iteration equation is:

$$
\langle MATH_3 \rangle
$$

where fₖⁿ and fₖ₊₁ⁿ are the reconstructed images after the k-th and (k+1)-th iterations, respectively, and sⁿ is the sensitivity matrix, which represents the detection efficiency of the detector for photons in direction n. tₘₙ is the system response matrix, representing the likelihood that the event originated from image pixel n.

4. Experimental Results and Discussion

To evaluate the imaging performance of this system, this study used multiple gamma point sources for testing. Projection data for these sources were collected at various positions within the field of view to perform single-source imaging experiments. Furthermore, detection events from point sources at different positions were combined to simulate imaging scenarios with multiple radioactive sources. The system employs a spherical coordinate system to describe the spatial distribution of radioactive sources, where the azimuthal angle φ denotes the projection angle of the radioactive source onto the equatorial plane and the polar angle θ is its inclination angle relative to the vertical axis. This configuration ensures effective coverage of the entire 4π steradian space, satisfying the geometric requirements for radiation field reconstruction.

4.1 Detection Efficiency and Energy Resolution

The detection efficiency of the Compton imaging system refers to the ratio of valid events available for imaging to the total number of detected gamma-ray events. In this experiment, a ¹³⁷Cs source with an activity of 2.35×10⁶ Bq was placed 1 meter away from the detector, and a total of 1.0×10⁶ events were collected. After subtracting environmental background, the net gamma-ray counts were 899,075. Following position reconstruction, energy reconstruction, and imaging event selection, 4,604 valid two-point events meeting the Compton imaging criteria were selected. The four-layer Compton imaging system exhibits approximately double the detection efficiency of its two-layer counterpart under identical experimental conditions. This improvement is primarily attributed to the multi-layer detector structure, which more effectively suppresses the escape of scattered photons, thus enhancing valid event capture capability.

Gamma-ray imaging systems require not only high detection efficiency but also superior energy resolution as critical performance metrics. Since the energy deposition for each pixel in the detector is used to calculate the Compton scattering angle, it is necessary to separately evaluate the energy resolution of each detector module within the omnidirectional Compton camera. Figure 6(a) shows the energy resolution distribution for all detectors. The distribution demonstrates that the energy resolution of all detectors in the omnidirectional Compton camera is highly consistent, which is crucial for ensuring imaging performance. To further evaluate the overall energy response of the system, the total energy spectrum was also tested. Figure 6(b) shows the total energy spectrum of ¹³⁷Cs following energy calibration. The full width at half maximum (FWHM) of the full-energy peak (662 keV) was 41 keV, and the total energy resolution was calculated to be 6.2% at 662 keV. This energy resolution is excellent for Compton imaging, allowing the Compton camera to accurately measure and distinguish radioactive sources with different energies.

4.2 Radioactive Source Imaging

In the single-point source imaging experiment, a ¹³⁷Cs source was placed at different positions 1 meter away from the detector, at the same height as the center of the field of view. The placement of the radioactive source at different positions enables evaluation of the system's capability for omnidirectional detection and imaging. Figure 7 FIGURE:7 shows the reconstructed image of the ¹³⁷Cs point source obtained using the MLEM algorithm after 10 iterations. The image clearly shows that the source is accurately localized and exhibits high angular resolution. To evaluate the angular resolution, a cross-sectional profile was extracted from Figure 7(a) along the direction of φ = 90°, as shown in Figure 7(b). Based on Gaussian fitting, the angular resolution of the imaging system was found to be approximately 12° (FWHM). Compared with coded aperture imaging, Compton imaging exhibits relatively poorer angular resolution due to combined effects from detector position resolution, Doppler broadening, and energy resolution.

Figure 7(c) shows the reconstructed image of two ¹³⁷Cs point sources placed at different azimuths. When the two ¹³⁷Cs point sources are placed sufficiently far apart, they can be clearly separated at the correct azimuth, demonstrating the system's multi-source localization capability. More importantly, the image demonstrates that the system, with its four-layer detector structure, achieves omnidirectional Compton imaging, thereby overcoming the field-of-view limitations of conventional Compton imaging systems. Figure 7(d) shows the reconstructed image of ¹³⁷Cs and ⁶⁰Co point sources placed at different azimuths. The hybrid imaging result clearly resolves both point sources with different energies and demonstrates good angular resolution. Moreover, when there is a large activity difference between the two point sources, excessive iterations will enhance the incident direction of the dominant source while causing the weaker source to disappear during iteration, resulting in biased localization. Therefore, the number of iterations should be limited, typically not exceeding 20 iterations.

4.3 Discussion

An omnidirectional Compton camera with a square ring-shaped structure offers significant improvements in detection efficiency, energy resolution, and azimuthal sensitivity compared to traditional Compton cameras with two planar detectors. In this study, to simulate far-field imaging and considering the size of the camera, radioactive sources were placed 1 m away from the detectors. The data were acquired and reconstructed to demonstrate the effectiveness of the novel Compton geometry. The experimental results show that the proposed omnidirectional Compton camera performs well in both single-point and multi-point source imaging scenarios.

The four-layer detector structure effectively suppresses the escape of scattered photons. By capturing more scattered photons, the system significantly enhances the number of valid events available for image reconstruction, thereby improving the quality of the reconstructed image. Additionally, the modular design and serial readout circuit ensure that signals from each detector module are accurately captured and processed, which is crucial for enhancing the energy resolution of the system. Experimental results show that all detector modules exhibit excellent and consistent energy resolution, with a total energy resolution of 6.2% (FWHM) at 662 keV. The enhanced energy resolution meets the requirements for distinguishing common radioactive sources, ensuring the ability of the Compton camera to image multiple point sources of different types.

In this study, the omnidirectional Compton camera with a four-layer detector structure eliminates the field-of-view limitations inherent in traditional two-layer Compton cameras. To validate its omnidirectional imaging capability, experiments were conducted by placing single-point sources, multi-point sources of the same type, and multi-point sources of different types at different azimuths around the detectors. The results show that the reconstructed image can accurately localize source positions with an angular resolution of 12°, and regardless of the direction of the radioactive source, the system successfully reconstructs the images. This confirms that the Compton camera has omnidirectional detection and imaging capabilities. Future research could further optimize detector structures to enhance angular resolution, as well as develop neural network methods to address imaging challenges in the presence of both strong and weak sources.

5. Conclusions

In this work, an omnidirectional Compton camera based on a square ring-shaped structure was developed, effectively resolving the field-of-view limitations of traditional imaging systems. The detector component adopted a modular design. Considering both energy resolution and position resolution, an 8×8 CsI(Tl) detector array was selected for the planar detector and constructed into four-layer detectors. To address the issue of excessive electronics channels, a serial readout circuit was implemented. This reduced the number of readout channels to only 32, effectively lowering system complexity while maintaining energy resolution. Through position and energy reconstruction combined with event selection, valid events were successfully selected for imaging, thereby improving the imaging quality of the Compton camera.

Experimental results demonstrated that the system provided a 360° horizontal field of view and achieved nearly twice the detection efficiency of conventional systems. In terms of performance, it exhibited a total energy resolution of 6.2% at 662 keV and an angular resolution of 12° for gamma-ray imaging using the MLEM algorithm. In addition, the system demonstrated the capability to distinguish different radioactive sources. This omnidirectional Compton camera has established a reliable technical foundation for radioactive source localization in complex environments.

References

[1] Y.S. Kim, J.H. Kim, H.S. Lee et al., Development of Compton imaging system for nuclear material monitoring at pyroprocessing test-bed facility. J. Nucl. Sci. Technol. 53(12), 2040-2048 (2016). https://doi.org/10.1080/00223131.2016.1199333
[2] K. Vetter, R. Barnowski, J.W. Cates et al., Advances in nuclear radiation sensing: Enabling gamma-ray vision. Sensors 19(11), 2541 (2019). https://doi.org/10.3390/s19112541
[3] X. Liu, P. Feng, Z.Y. Yao et al., Development of Imaging System of Array Compton Camera for Nuclear Decommissioning Scenarios. Acta Opt. Sin. 42(24), 2411002 (2022). https://doi.org/10.3788/AOS202242.2411002
[4] Z.Y. Yao, Y.G. Yuan, J. Wu et al., Rapid Compton camera imaging for source terms investigation in the nuclear decommissioning with a subset-driven origin ensemble algorithm. Radiat. Phys. Chem. 197(3), 110133 (2022). https://doi.org/10.1016/j.radphyschem.2022.110133
[5] Y. Sato, Radioactivity Estimation of Radioactive Hotspots Using a Compton Camera and Derivation of Dose Rates in the Surrounding Environment. Appl. Radiat. Isot. 212, 111421 (2024). https://doi.org/10.1016/j.apradiso.2024.111421
[6] V. Babiano-Suárez, J. Balibrea-Correa, I. Ladarescu et al., A Computer-Vision Aided Compton-Imaging System for Radioactive Waste Characterization and Decommissioning of Nuclear Power Plants. Nucl. Instrum. Methods Phys. Res. A 1076, 170449 (2025). https://doi.org/10.1016/j.nima.2025.170449
[7] Z.Y. Yao, Y.S. Xiao, J.Z. Zhao, Dose reconstruction with Compton camera during proton therapy via subset-driven origin ensemble and double evolutionary algorithm. Nucl. Sci. Tech. 34(4), 59 (2023). https://doi.org/10.1007/s41365-023-01207-1
[8] W. Lee, T. Lee, M. Jeong et al., Evaluation of dual γ-ray imager with active collimator using various types of scintillators. Appl. Radiat. Isot. 69(10), 1560-1567 (2011). https://doi.org/10.1016/j.apradiso.2011.07.003
[9] M. Jeong, B. Van, B. T. Wells et al., Scalable gamma-ray camera for wide-area search based on silicon photomultipliers array. Rev. Sci. Instrum. 89(3), 033106 (2018). https://doi.org/10.1063/1.5016563
[10] H3D, Inc. H100 Gamma-ray Imaging Spectrometer. https://h3dgamma.com/H100Specs.pdf, Accessed: 2025-04. [EB/OL]. Available:
[11] J.P. Zhang, X.Z. Liang, J.L. Cai et al., Prototype of an Array SiPM-Based Scintillator Compton Camera for Radioactive Materials Detection. Radiat. Detect. Technol. Methods 3, 17 (2019). https://doi.org/10.1007/s41605-019-0095-1
[12] Y.F. Hu, P. Fan, Z.L. Lyu et al., Design and performance evaluation of a 4π-view gamma camera with mosaic-patterned 3D position-sensitive scintillators. Nucl. Instrum. Methods Phys. Res. A 1023, 165971 (2022). https://doi.org/10.1016/j.nima.2021.165971
[13] J. Kataoka, A. Kishimoto, T. Fujita et al., Recent Progress of MPPC-Based Scintillation Detectors in High Precision X-Ray and Gamma-Ray Imaging. Nucl. Instrum. Methods Phys. Res. A 784, 248-254 (2015). https://doi.org/10.1016/j.nima.2014.11.004
[14] R. K. Parajuli, M. Sakai, R. Parajuli et al., Development and Applications of Compton Camera—A Review. Sensors 22(19), 7374 (2022). https://doi.org/10.3390/s22197374
[15] J.P. Zhang, X. Xiao, Y. Chen et al., A Portable Three-Layer Compton Camera for Wide-Energy-Range Gamma-ray Imaging: Design, Simulation and Preliminary Testing. Sensors 23(21), 8951 (2023). https://doi.org/10.3390/s23218951
[16] Z.Q. Yuan, D.Y. Xue, H. Yang et al., A novel Compton camera with an annular absorber for enhancing the imaging efficiency in regions directly ahead: A simulation study. J. Nucl. Sci. Technol. 60(10), 1208-1218 (2023). https://doi.org/10.1080/00223131.2023.2239135
[17] X.Z. Liang, L. Shuai, Y.T. Liu et al., Coded aperture and Compton imaging capability of spherical detector system design based on GAGG scintillators: A Monte Carlo study. Nucl. Instrum. Methods Phys. Res. A 1044, 167503 (2022). https://doi.org/10.1016/j.nima.2022.167503
[18] R. He, X.Y. Niu, Y. Wang et al., Advances in Nuclear Detection and Readout Techniques. Nucl. Sci. Tech. 34(11), 157 (2023). https://doi.org/10.1007/s41365-023-01359-0
[19] B.L. Zhu, Z.H. Wang, J.L. Zhang et al., Design and Implementation of a Portable High-Performance Gamma-Ray Camera. IEEE Trans. Nucl. Sci. 68(6), 1309-1318 (2021). https://doi.org/10.1109/TNS.2021.3078302
[20] X.Y. Li, J. Yang, G.Q. Zeng et al., Identification Method of Detector Position Based on Serial Readout Circuit. Nucl. Tech. 48(6), 060008 (2025). https://doi.org/10.11889/j.0253-3219.2025.hjs.48.240345
[21] G.Q. Zeng, J. Yang, T.Y. Hu et al., Baseline restoration technique based on symmetrical zero-area trapezoidal pulse shaper. Nucl. Instrum. Methods Phys. Res. A 858, 57-61 (2017). https://doi.org/10.1016/j.nima.2017.03.049