RT Journal Article
T1 Efficient Real-Time Isotope Identification on SoC FPGA
A1 Guerrero Morejón, Katherine
A1 Hinojo Montero, José María
A1 Jiménez Sánchez, Jorge
A1 Rocha Jácome, Cristian
A1 González Carvajal, Ramón
A1 Muñoz Chavero, Fernando
AB Efficient real-time isotope identification is a critical challenge in nuclear spectroscopy, with important applications such as radiation monitoring, nuclear waste management, and medical imaging. This work presents a novel approach for isotope classification using a System-on-Chip FPGA, integrating hardware-accelerated principal component analysis (PCA) for feature extraction and a software-based random forest classifier. The system leverages the FPGA’s parallel processing capabilities to implement PCA, reducing the dimensionality of digitized nuclear signals and optimizing computational efficiency. A key feature of the design is its ability to perform real-time classification without storing ADC samples, directly processing nuclear pulse data as it is acquired. The extracted features are classified by a random forest model running on the embedded microprocessor. PCA quantization is applied to minimize power consumption and resource usage without compromising accuracy. The experimental validation was  conducted using datasets from high-resolution pulse-shape digitization, including closely matched isotope pairs (12C/13C, 36Ar/40Ar, and 80Kr/84Kr). The results demonstrate that the proposed SoC FPGA system significantly outperforms conventional software-only implementations, reducing latency while maintaining classification accuracy above 98%. This study provides a scalable, precise, and energy-efficient solution  for real-time isotope identification.
PB MDPI
SN 1424-8220 (electrónico)
YR 2025
FD 2025
LK https://hdl.handle.net/10272/27486
UL https://hdl.handle.net/10272/27486
LA eng
NO Guerrero-Morejón, K., Hinojo-Montero, J. M., Jiménez-Sánchez, J., Rocha-Jácome, C., González-Carvajal, R., & Muñoz-Chavero, F. (2025). Efficient Real-Time Isotope Identification on SoC FPGA. Sensors, 25(12), 3758. https://doi.org/10.3390/s25123758
NO This research was supported by grant CPP2022-009904, funded by MCIN/AEI/ 10.13039/501100011033 and the “ERDF A way of making Europe”, and by grant PDC2023-145828-C21, funded by MCIN/AEI/10.13039/501100011033 and “ERDF A way of making Europe”.
DS Repositorio Institucional de la Universidad de Huelva
RD 14 jul 2026