Naasih Delvi
Research Mentor: Jordan Noey
Mentor Department: Nuclear Engineering and Radiological Sciences, Engineering
Author(s): Naasih Delvi, Jiahui Han, Jordan Noey, Kimberlee Kearfott
Session: Session 1 (9:00 AM – 9:50 AM)
Presentation Type: Poster 91
Abstract
Thermoluminescent dosimeters are well-known tools that provide information on integrated dose. When heated, they produce glow curves that quantify absorbed radiation doses, and researchers can interpret this data using models that describe physical processes. Software programs are a fast and convenient way to interpret glow curve data, but individual researchers often require flexibility to customize their scientific approach. Many existing software methods necessitate manual peak approximation or significant code refactoring, which is too inefficient for many researchers. This project focuses on improving accessibility by building an expanded web interface for the glow curve analysis software, designed and repackaged so that people with varying levels of technical experience can still conduct in-depth analysis. The web application utilizes a Python backend that incorporates a C++ executable, complemented by a frontend built with HTML, CSS, Flask and Jinja2. This work specifically aims to preserve the efficiency of automated methods while improving the accuracy of peak identification. The implementation is divided into two stages: firstly, the verification of the peak fitting algorithm across various conditions, including low doses and varying heating rates; secondly, a systematic redesign of the peak detection component, adapting the hard-coded initial guess input in the stochastic gradient descent by incorporating the first- and second-order derivative method, which allows for flexible adaptation of the initial guesses. These refinements will result in figures of merit that are systematically lower and agree with literature. The goal is to transition the glow curve analysis software to an open-source repository, providing a transparent and accessible package for curve fitting that other researchers will benefit from.
