Speaker
Description
Gamma-ray spectrometry remains a cornerstone technique in nuclear science, offering precise information on radionuclide identification and quantification. However, conventional spectrum analysis methods based on peak fitting, background subtraction, and energy calibration are often time-consuming and sensitive to operator bias, especially in complex or noisy spectra. In this context, the integration of Artificial Intelligence (AI) and Deep Learning (DL) represents a significant advancement toward automation, accuracy, and high-throughput analysis.
This work explores the application of deep neural network architectures, particularly Convolutional Neural Networks (CNNs) and Autoencoders, for automated processing and interpretation of gamma-ray spectra acquired from HPGe detectors. The developed models are trained on both simulated and experimental data to perform key analytical tasks such as peak detection, nuclide identification, and activity quantification. Preliminary results demonstrate that AI-based models can accurately identify overlapping peaks, reduce background interference, and provide rapid isotope classification, outperforming traditional analytical algorithms in robustness and speed.
The proposed AI framework offers promising perspectives for real-time monitoring, nuclear safeguards, and environmental radioactivity assessment. It also contributes to the ongoing evolution of intelligent spectrometry systems capable of self-learning and adaptive calibration. Future work will focus on integrating these models with laboratory instrumentation for fully automated analysis workflows and expanding their applicability to other nuclear detection systems.