Nuclear energy is a clean and popular form of energy, but leakage and loss of nuclear material pose a threat to public safety. Radiation detection in public spaces is a key part of nuclear security. Common security cameras equipped with complementary metal oxide semiconductor (CMOS) sensors can help with radiation detection. Previous work with these cameras, however, required slow, complex frame-by-frame processing. Building on the previous work, we propose a nuclear radiation detection method using convolution neural networks (CNNs). This method detects nuclear radiation in changing images with much less computational complexity. Using actual video images captured in the presence of a common Tc-99m radioactive source, we construct training and testing sets. After training the CNN and processing our test set, the experimental results show the high performance and effectiveness of our method.

NaI(Tl) spectrometry is widely used in many areas for gamma-ray detection. It can be used for both activity measurement and isotope identification. However, the light generation of NaI(Tl) and photomultiplier tube (PMT) is sensitive to temperature changes, which lead to spectrum shift. Several calibration methods have been proposed, but they are complicated. In this article, we developed an NaI(Tl) spectrometry for indoor monitoring and studied the temperature influences from 5 °C to 35 °C using a ²² Na source. We found that the severity of the temperature-dependent spectrum shift can be reduced by using fewer integration points (i.e., short integration time) for energy calculation. The spectrum shift can be reduced from 18.1% to 3.3% for 511 keV and from 18.0% to 3.7% for 1274 keV by using an integration time of 385 ns instead of 1.23 μs . In conclusion, the spectrum shift phenomenon can be reduced by reducing integration time. Thus, the indoor monitoring NaI(Tl) spectrometry might work properly for isotope identification without a complicated calibration procedure.

Purpose: Accurate patient-specific absorbed dose estimation is important for radionuclide therapy, but its routine clinical use is limited by the high computational cost of full Monte Carlo simulations. Voxel S value (VSV)-based convolution methods are efficient, but their accuracy may decrease in heterogeneous media, especially for short-range electrons. This study presents a practical hybrid dose calculation framework for I-131 therapy to balance accuracy and efficiency. Methods: In the proposed framework, electron transport was simulated using Monte Carlo methods, whereas photon dose was calculated by convolution with pre-calculated VSV kernels. Total absorbed dose was obtained by voxel-wise summation of electron and photon contributions. Photon VSV accuracy was first validated against MIRD Pamphlet No. 17 data. The hybrid method was then evaluated in heterogeneous phantoms with different material densities and in a clinical post-therapy SPECT/CT dataset. Results: The photon VSV agreed well with the MIRD Pamphlet No. 17 reference data. In heterogeneous media, the hybrid method improved dose estimation in low-density regions compared with the conventional VSV approach and showed close agreement with full Monte Carlo simulation. In the patient study, computation time was reduced by about sevenfold while improved dosimetric accuracy was maintained in regions where the conventional VSV method showed larger deviations. Conclusions: The proposed hybrid framework provides an efficient strategy for voxel-level dose estimation in I-131 therapy, improving accuracy in heterogeneous media while substantially reducing computation time. It may support personalized internal dosimetry when integrated with quantitative imaging and time-integrated activity estimation.