Introduction

      Gamma-ray imaging technology plays a crucial role in various fields, including astronomical observation, medical applications, nuclear power plant decommissioning, and nuclear security. Gamma-ray imaging methods are generally of three types: those using physical collimators to limit the gamma-ray direction (such as pinhole cameras and single-photon emission computed tomography devices), those employing electrical collimators (such as Compton cameras and positron emission tomography devices), and those that spatially or temporally modulate the field using shielding (such as coded masks and rotating modulation). Each method is selected based on its specific application, considering factors such as weight, angular resolution, and field-of-view (FOV).

At the Fukushima Daiichi Nuclear Power Station decommissioning site, which is our focus, there’s a need for gamma-ray imaging technology that is compact and lightweight enough to be mounted on autonomous exploration robots, has a wide FOV and can operate in high-dose-rate environments. Coded masks and Compton cameras, although small and light, suffer from limitations such as a narrow FOV, susceptibility to background noise, or poor angular resolution and unsuitability for high-dose environments, respectively. Therefore, we developed a novel gamma-ray imaging technology that weighs <600 g, has a cubic form measuring 8.6 cm on each side, offers a 4π FOV, has an angular resolution of approximately 5°, and can operate in environments with radiation levels up to 100 mSv/h with only eight readout channels.

Coded Cube Camera for Gamma-ray (C3G)   

This section briefly explains the principle of our proposed method and describes the prototype we built. The proposed method can be considered as a three-dimensional (3D) extension of coded-mask cameras. Figure 1 illustrates the principles of the coded-mask camera and the proposed method, with an analogy determining the direction of the sun. The coded mask camera infers the direction of the sun from a 2D shadow pattern on the ground and the shape of clouds, corresponding to the coded mask and position-sensitive detector, respectively. Since the incidence direction of gamma rays determines the shadow pattern projected by the coded mask onto the position-sensitive detector, the measured shadow pattern identifies the incidence direction of the gamma ray.

This technique does not require heavy shielding, achieves high angular resolution and supports imaging at relatively high-count rates without the need for coincidence. However, its reliance on gamma rays that pass through the coded mask makes it sensitive to background noise and limits its FOV. The proposed method uses the 3D pattern of light streaks in the atmosphere. In this analogy, the shielding cubes and detectors correspond to the clouds and atmosphere, respectively. This approach allows for a 4π expansion of the FOV because the shielding cubes modulate all directions of the gamma rays.

We used this method to develop a new gamma-ray imager, the C3G. Figure 2 shows the prototype of C3G, which comprises 18 lead shielding cubes and 8 GAGG (Ce) scintillator cubes.

Experiment

To test the performance of the C3G prototype, we conducted a comparative experiment using a ¹³⁷Cs source with C3G and a Compton camera. The results indicate that C3G yielded clear images with a wide FOV (Figure 3). The image reconstruction was conducted using machine learning. In this experiment, C3G achieved an angular resolution of approximately 5° using eight readout channels. The angular resolution exceeds that of Compton cameras and is comparable to that achieved by coded-mask cameras. It weighs less than 600 g and has a 4π FOV with only eight readout channels. Typically, gamma-ray imagers require hundreds to thousands of readout channels, which involve complex signal processing and expensive and specialized circuits. However, the proposed method allows for simple signal processing, enabling the development of flexible and versatile systems. In the future, the arrangement of shield cubes will be optimized, and the performance of reconstruction algorithms will be improved to create a more practical imager.

Yoshiharu Kitayama: Doctoral student at Tohoku University and an employee at JAEA, can be reached by E-mail at: [email protected]