Context Prostate cancer is the most common cancer in 112 countries, accounting for 15% of all cancer cases in men. The number of diagnoses is expected to increase from 1.4 million in 2020 to 2.9 million by 2040 (James et al., 2024). Today, prostate cancer diagnosis is quite early and precise, particularly with advancements in PSA blood testing and MRI technology. However, even when tumors are small and well-identified, conventional treatments affect the entire gland, leading to significant side effects. A more recent approach involves developing focal treatment methods that target only the tumor area (Graff et al., 2018). The aim is to treat the cancer effectively while minimizing side effects by preserving healthy tissue. This type of treatment comes with unique challenges. While global treatments ensure complete coverage of the tumor, focal treatments demand precise planning and execution. Without this precision, there's a risk of inadequately treating the cancer, which could lead to recurrence. Therefore, developing advanced planning systems to determine the optimal treatment positions and intensities is essential for improving the quality and effectiveness of focal treatments. Main objective In the past, we developed a planning system for low-dose focal brachytherapy for prostate cancer (Mountris et al., 2019; Villa et al., 2022). To explore other treatment modalities such as cryotherapy (Fig. 1-a), microwave therapy (Fig. 1-b) and electroporation (Fig. 1-c), mathematical models that define the treated area based on physical parameters (tissue type, energy, treatment time, etc.) need to be created. These models are crucial for estimating precise treatment criteria, such as tumor coverage and impact on organs at risk. These criteria are then used to define a cost function that, along with an objective function, enables the optimization algorithm to converge toward a clinically acceptable solution. For this, we use the simulated annealing method for its non-convex and multiparametric search capabilities, which help avoid local minima. The goal of the internship will be to create mathematical and algorithmic models for different treatments using real data and literature. These models must be precise and efficient, and they will be tested in our planning software. For each treatment modality, the cost function, with optimized criteria estimates in terms of computation time and complexity, must also be developed. The aim is to ensure rapid convergence suitable for a clinical context. The treatment plan results will be presented and evaluated by a urologist experienced in these treatments.

Master2 in computer science, applied mathematics, physics, biomedical engineering. Autonomy, open-mindedness and motivation.