Previsão Computacional Personalizada de Adenocarcinomas da Próstata

This interdisciplinary project addresses a timely, unresolved medical problem: the personalization of the clinical management of PCa to reduce overtreatment and undertreatment. We propose to build and validate a personalized, organ-scale mathematical model of PCa to forecast tumor growth and assist physicians in diagnosis and clinical-decision making. Multiparametric magnetic resonance (MR) imaging has been providing a wealth of data to describe the tumor morphology, architecture and behavior [M]. Given the increasing use of mpMRI data to inform clinical decision-making in PCa [R], it is timely to validate the predictive capability of state-of-the-art tumor forecasting models for PCa cases against longitudinal patient-specific data [3]. Recently, mechanistic and reproducible computational methods have enabled the personalized prediction of clinical outcomes and the design of optimal therapies, e.g., for breast and brain cancers [F,G,H]. This new approach, termed computational oncology, combines the use of mathematical models accounting for key physical and biological mechanisms with computer simulations to forecast tumor growth. Personalization of these predictions relies on the parameterization of the model with longitudinal clinical and imaging data from each patient and on simulating tumor growth over the actual anatomy of the patient’s affected organ. In this project we will extend the clinic use of these predictive strategies to PCa, which presents the possibility for validation through a direct comparison with surgical samples’ histologies. Furthermore, the use of isogeometric analysis (IGA)[Q] to integrate the model equations, a powerful generalization of the Finite Element Method, pushes forward the state of the art of this approach, and will address accurately and efficiently the computational challenges of PF models in patient-specific prostate anatomies. Finally, we will explore the capabilities for the validated model to predict and compare a personalized PCa growth after focal therapy, external beam radiation therapy and brachytherapy. Importantly, the use of computational oncology in PCa prediction has striking advantages over other trending model-naïve approaches, such as machine learning [R] or computer-aided diagnosis systems [S]. These rely on statistical methods to predict, respectively, the probability of tumor-related features or endpoints at certain times and the identification of patterns in imaging data. However, computational oncology (1) accounts for the biological and physiological phenomena ultimately underlying cancer dynamics; (2) enables the physically quantitative assessment of tumor growth over time within the tissue, facilitating clinical decision-making; and (3) it is based on mechanistic models, so it is fully reproducible and interpretable patient-wise. This project brings together a multidisciplinary team of excellence, tailored to successfully address its challenging tasks. The collaboration with clinicians and the use of patient data guarantees the applicability of findings in the clinic, potentiating a disruptive improvement in the clinical management of PCa. Additionally, this project provides knowledge transfer between specialists and the society, has a strong focus in training new scientists in an outstanding interdisciplinary environment, and presents an opportunity to foster this multidisciplinary network of excellence, which will extend far beyond the project lifespan.