【摘要】 互补金属氧化物半导体（CMOS）有源像素传感器（APS）作为新型高灵敏度光电探测器件，在辐射成像领域展现出广泛的应用潜力。提出一种基于CMOS APS图像像素值的β射线定量探测模型。设计⁶³Ni β源辐照实验与蒙特卡罗仿真实验，对CMOS APS在不同增益下的β射线辐射响应事件特征，不同积分时间下的暗图像平均像素值、能量沉积，以及损伤模式和室温退火效应进行分析。实验结果表明：通过双阈值约束增益优化法得出β射线测量的最佳增益为43 dB;CMOS APS随着辐照时间增加，损伤情况不断加剧，室温退火效应导致其损伤趋势为非线性变化；时间差分滤波法能够良好地去除噪声，并有效保留β响应事件，其保留率达到99.75%，且信噪比相较于降噪前提升至5.17倍；当传感器处于5.625 ms以下的低积分时间区域时，计算能量沉积与仿真能量沉积的匹配度较高，而在高于5.625 ms的高积分时间区域时传感器像素出现电荷饱和效应，误差显著增大；基于分段指数衰减校正后的像素值-剂量探测模型能够实现对β射线吸收剂量的量化。更多还原

【Abstract】 Objective β-radiation monitoring plays a significant role in fields such as nuclear safety, environmental protection, radioactive waste management, and the application of medical isotopes. Compared with α and γ rays, β rays are more difficult to detect due to their weak penetrating power and wide energy spectrum distribution. Traditional β detectors are often limited by low spatial resolution, insufficient sensitivity and difficulty in miniaturization. With the development of commercial complementary metal oxide semiconductor（CMOS） image sensor technology, CMOS active pixel sensors（APS） have shown broad prospects in the field of low-dose radiation detection due to their high integration, low power consumption, low cost and excellent imaging capabilities. However, the quantitative relationship between the energy deposition of β particles in CMOS and the gray-scale response of pixels remains unclear, and the noise caused by radiation damage also compromises the detection accuracy. Therefore, this paper aims to systematically study the response characteristics of CMOS APS under β-irradiation conditions and construct a reliable correlation model between pixel grayscale values and absorbed doses. The research results provide a new path for achieving low-cost and high-resolution β-radiation detection and offer theoretical support for the extended application of CMOS sensors in complex radiation environments.Methods In this paper, the CMOS APS image sensor MT9P031 is selected as the experimental object, and a ⁶³Ni planar β source with a maximum energy of 66.9 keV is used for irradiation. Images are collected in a dark environment, and the radiation response events caused by β particles in the images are extracted. They are identified and integrated based on their gray-scale level and morphological characteristics. During the statistical process, a double-threshold constraint method based on response event saturation and allowable degree is adopted to determine the optimal gain in the detection process. The absorbed dose of the sensor is calculated by integrating the conversion gain and the total gray-scale value of the image response events. Meanwhile, a CMOS structure Geant4 simulation model is constructed to simulate the energy deposition process of β particles inside it, and the energy deposition results obtained from image processing are verified. The piecewise exponential function is adopted to correct the error caused by the charge saturation effect of the image sensor pixels due to the high integration time of the sensor. In addition, the characteristic laws of radiation noise generated by β rays are analyzed. The time-difference radiation noise filtering method is adopted to remove the unresponsive pixel noise caused by damage in the image and eliminate the error interference caused by damage to the greatest extent.Results and Discussions The research shows that the β radiation response events are randomly distributed on the sensor image and are characterized by the central pixel approaching saturation accompanied by an increase in the values of the peripheral pixels（Fig. 2）, and the response intensity is significantly affected by the gain and integration time（Fig. 3）. By introducing a dual-threshold constraint method based on response event saturation and allowable degree, the optimal gain is clearly defined as 43 dB, which ensures signal amplification while avoiding oversaturation distortion（Fig. 4）. Further analysis reveals that the dark signals induced by β irradiation during the irradiation process have significant accumulation effects and nonlinear annealing characteristics（Fig. 8）, especially showing a reverse enhancement trend after 42 h of annealing（Fig. 9）. A time-differential filtering algorithm is proposed, which effectively suppresses radiation noise while retaining β response events（Table 2, Fig. 11）. Compared with traditional filtering methods, the signal-to-noise ratio is improved by more than five times, and its advantages in single-frame image noise reduction and signal fidelity are verified by comparing with other methods. Using the sensor conversion gain k, the energy deposition of β rays in CMOS APS is calculated, ultimately enabling the determination of the absorbed dose in the CMOS APS. Meanwhile, a Geant4 simulation model is constructed to validate the energy deposition calculated experimentally, and a piecewise exponential correction function of“ pixel value-dose” is proposed, which can correct the error caused by the charge saturation effect of the image sensor pixels under high integration time（Fig. 13）, and solve the distortion problem in the strong signal region, achieving accurate mapping from pixel values to doses（Fig. 14）.Conclusions This study verifies that CMOS APS exhibits excellent detection and quantitative capabilities under lowdose β radiation conditions. Through experimental and simulation-based verification, we propose a β-ray detection method with high accuracy and stability. Combined with temporal differential filtering, the radiation noise generated during the extraction of β-radiation response events is effectively suppressed, enhancing measurement robustness. The research results show that CMOS APS has broad application potential in the field of low-cost and high spatial resolution β radiation measurement, and can be used in environmental safety, radiation pollution monitoring, and the development of portable dosimeters, among others.更多还原