Thesis Defense: Analyzing the complexity of radiomics: Approaches for simplified modeling

Description:

Speaker: Mahdi Ait Lhaj Loutfi, M. Sc. student in Computer Science, Department of Computer Science, Faculty of Science, Université de Sherbrooke

Abstract: Radiomics is an emerging discipline that uses quantitative data extracted from medical images to develop predictive models. Despite its growing adoption in various research contexts, radiomics faces several challenges, including the high dimensionality of radiomic feature sets, the variability of radiomic feature types, and potentially high computational demands. These challenges highlight the need for an effective method to identify the simplest and most efficient features for a specific clinical problem.

In this thesis, we will address the concept of radiomic complexity levels, defined by the computational steps required to extract a type of feature. We will then test the hypothesis that there is an optimal level of complexity for a given clinical problem. Consequently, we will identify these optimal levels for various clinical applications. This question is central to simplifying radiomic approaches. Through the analysis of five datasets with different imaging modalities to predict various clinical outcomes, we will extract radiomic features for each complexity level and evaluate their performance to identify the optimal level, defined by the most performant level with the minimum number of features. We will then validate our hypothesis of the existence of an optimal radiomic level for a given clinical problem and confirm the effectiveness of our methodology in determining this level. We will then demonstrate how this identification can lead to performance improvements through in-depth analyses. Additionally, we will present the tools developed during this project, which integrate our methodology and make radiomic analyses accessible to professionals from various disciplines. We will also highlight the synergy between computer scientists and healthcare professionals, essential for radiomic applications.

Jury member, president François Rheault, Professor, Department of Computer Science, Faculty of Science, Université de Sherbrooke

Jury member, research director: Martin Vallières, Professor, Department of Computer Science, Faculty of Science, Université de Sherbrooke

Jury member, research co-director: Martin Lepage, Professor, Faculty of Medicine and Health Sciences, Department of Nuclear Medicine and Radiobiology, Université de Sherbrooke

Jury member, external evaluator: An Tang, Professor, Department of Radiology, Radiation Oncology, and Nuclear Medicine, Université de Montréal

All interested persons are cordially invited.

Teams link: https://bit.ly/46vSRyd