Peak-Based Nuclide Identification in HPGe $γ$-Spectrometry with Machine Learning and SHAP

2026-06-12 · Source: Machine Learning · Field: Science & Research — Physical Sciences & Chemistry, Artificial Intelligence & Machine Learning, Research Methodology & Innovation · Depth: Expert, quick

Summary

A new machine learning approach significantly improves nuclide identification (NID) in High-purity germanium (HPGe) gamma-spectrometry, a process traditionally requiring extensive expert analysis. Researchers implemented ML models that map photopeaks, carefully fitted by analysts, to NID results for experimental spectra containing various isotopic combinations. These models were trained and assessed using a library of 65 isotopes. The top-performing ML model achieved an F1 score of 0.97, substantially surpassing the 0.84 F1 score of conventional software when evaluated against the identical 65-isotope library. Furthermore, Shapley Additive Explanations (SHAP) were employed to illustrate the most important input features for model predictions, confirming that the models utilize physically relevant photopeaks for isotope identification. This automation can enhance the initial radionuclide suggestions for analysts.

Key takeaway

For HPGe spectrometry analysts seeking to accelerate nuclide identification, integrating machine learning models can significantly improve efficiency and accuracy. You should consider deploying these models to automate initial radionuclide suggestions, potentially reducing manual analysis time and enhancing subsequent quantification. This approach, validated by an F1 score of 0.97, offers a robust alternative to traditional software, which achieved 0.84.

Key insights

Machine learning models achieve superior nuclide identification in HPGe gamma-spectrometry, reaching an F1 score of 0.97, surpassing traditional software.

Principles

• ML automates expert-informed nuclide identification.
• Explainability validates ML's use of physical features.

Method

Machine learning models were implemented to map analyst-fitted photopeaks to nuclide identification results. These models were trained on experimental spectra from 65 isotopes, with SHAP used to explain feature importance.

In practice

• Enhance initial radionuclide suggestions.
• Drive effective subsequent quantification.

Topics

• HPGe Gamma-Spectrometry
• Nuclide Identification
• Machine Learning
• SHAP
• Photopeak Fitting
• Nuclear Data Analysis

Best for: Research Scientist, AI Scientist, Machine Learning Engineer

Related on AIssential

• SpecGP as a transformer-based model for predicting energy-adaptable structural spectra of glycopeptides — SpecGP uses a transformer architecture to predict glycopeptide spectra and retention times, enhancing isomeric discrimination and identification.
• Machine-Learning Emulation of Satellite Greenhouse Gas Retrievals: Stability over Time — Machine-learning emulators for GHG retrievals require time-aware training to maintain accuracy and stability over extended periods.
• MōLe-Λ: Learning the Coupled-Cluster Response State for Energies, Gradients, and Properties — MōLe-Λ efficiently predicts full CCSD response states by learning both right- and left-hand amplitudes, expanding accessible quantum chemistry properties.
• Integration of alternative fragmentation techniques into standard LC-MS workflows using a single deep learning model enhances proteome coverage — A unified deep learning model and integrated MS platform enhance proteome coverage by leveraging diverse fragmentation techniques.
• Auto-FP: An Experimental Study of Automated Feature Preprocessing for Tabular Data — Automating feature preprocessing for tabular data can be framed as HPO or NAS, with evolution-based algorithms performing best.
• Machine Learning-Based Pre-Test Risk Stratification for PCR-Confirmed Chlamydia Using Patient-Reported Data and Urine Biomarkers — Urine biomarkers offer a reliable, consistent signal for Chlamydia pre-test risk stratification, complementing patient-reported data.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.