Optimisation of radiation health monitoring through machine learning and data fusion

Radiation portal monitors can provide security and safety screening where radioactive material tracking is required. Combined walk-though systems can facilitate screening of metallic and radioactive materials simultaneously. Real-time portal radiation monitoring is well-suited to controlled environments; this can encompass scenarios where the level of radioactive contamination is known or unknown. In cases of natural disasters or human accidents, retrospective monitoring may be required. Both jewellery beads [1] and silica fibre [2] have showed promising preliminary results for delayed dosimetry [1,2]. This project will first compare personnel radiation readings from commercial equipment with that obtained through delayed in-situ material measurements, the latter obtained through thermoluminescence dosimetry experiments.  Research will then seek to optimise the information acquired from bead and fibre dosimetry through machine learning. This may require the creation of a convolutional neural network training dataset, using an augmented experimental seed method as developed for nuclear decommissioning [3]. Finally, data fusion approaches will assess the diagnostic information obtained when retrospective and real-time radiation monitoring methods are combined. The collective PhD scope will provide an informed assessment of delayed material dosimetry within the radiation health monitoring landscape. 

[1]           Fading and Residual Responses for Thermoluminescent Dosimetry of Silica Beads Irradiated Using a High-dose Electron-beam (2021) K Ley, S A Hashim, A Lohstroh, C Shenton-Taylor, D A Bradley, Radiation Physics and Chemistry doi.org/10.1016/j.radphyschem.2021.109366

[2]           Thermoluminescence Glow Curve Study of beta irradiated Ge doped core fibre with different dopant concentrations, (2022) C. Termsuk, S. Sweeney and C. Shenton-Taylor, Radiation Physics and Chemistry, doi.org/10.1016/j.radphyschem.2022.109974

[3]           An artificial neural network algorithm developed for shielded multi-isotope identification (2023) L Lee-Brewin, D. Read and C. Shenton-Taylor doi 10.1088/1748-0221/18/05/P05043

Open to UK nationals only*. You will need to meet the minimum entry requirements for our Physics PhD programme. Usually, you should hold a first-class, upper second class (or international equivalent) Bachelor’s or Master’s degree in physics or a related subject. Ideally you will have experience (or at least a strong interest) in nuclear physics, computational methods and machine learning. 

*and those with Aus/Canada/USA nationality.