<p dir="ltr">Human brain tumors present significant challenges due to their heterogeneous nature, known as intra-tumoral heterogeneity (ITH), which evolves over space and time, leading to treatment resistance and poor patient outcomes. Current diagnostic methods rely on pre-surgical imaging and single biopsy samples, providing only a partial understanding of the tumor microenvironment (TME) and often resulting in incomplete targeting of tumor mutations, leaving residual disease vulnerable to recurrence. Our hypothesis proposes a novel approach: utilizing multimodal and multiparametric <i>in vivo</i> imaging to map the cellular and molecular characteristics of the TME. By correlating imaging signatures with underlying somatic and genomic aberrations, we aim to develop a predictive model guiding personalized targeted therapies to effectively address the heterogeneity of brain tumors.</p><p dir="ltr">To achieve this goal, we designed, tested, and validated a predictive model through a pilot study using clinical MRI scans and one stereotactic biopsy sample. Subsequently, we optimized a multimodal and multiparametric imaging protocol including MRI and PET scans, to acquire comprehensive morphological, functional, and molecular data from the TME. Additionally, we established a detailed pipeline for subject recruitment, data collection, and post-processing to ensure the robustness and reliability of our model.</p><p dir="ltr">This innovative approach has the potential to overcome the limitations of current diagnostic methods by providing a comprehensive understanding of the TME using minimally-invasive imaging techniques. By correlating imaging data with ground truth pathology and genomics, this model will enhance brain tumor diagnosis and facilitate the implementation of targeted therapies, ultimately improving treatment response and patient outcomes.</p>

A better understanding of tumour heterogeneity is central for accurate diagnosis, targeted therapy and personalised treatment of glioblastoma patients. This thesis aims to investigate whether pre-operative multi-parametric magnetic resonance imaging (MRI) can provide a useful tool for evaluating inter-tumoural and intra-tumoural heterogeneity of glioblastoma. For this purpose, we explored: 1) the utilities of habitat imaging in combining multi-parametric MRI for identifying invasive sub-regions (I & II); 2) the significance of integrating multi-parametric MRI, and extracting modality inter-dependence for patient stratification (III & IV); 3) the value of advanced physiological MRI and radiomics approach in predicting epigenetic phenotypes (V). The following observations were made: I. Using a joint histogram analysis method, habitats with different diffusivity patterns were identified. A non-enhancing sub-region with decreased isotropic diffusion and increased anisotropic diffusion was associated with progression-free survival (PFS, hazard ratio [HR] = 1.08, P < 0.001) and overall survival (OS, HR = 1.36, P < 0.001) in multivariate models. II. Using a thresholding method, two low perfusion compartments were identified, which displayed hypoxic and pro-inflammatory microenvironment. Higher lactate in the low perfusion compartment with restricted diffusion was associated with a worse survival (PFS: HR = 2.995, P = 0.047; OS: HR = 4.974, P = 0.005). III. Using an unsupervised multi-view feature selection and late integration method, two patient subgroups were identified, which demonstrated distinct OS (P = 0.007) and PFS (P < 0.001). Features selected by this approach showed significantly incremental prognostic value for 12-month OS (P = 0.049) and PFS (P = 0.022) than clinical factors. IV. Using a method of unsupervised clustering via copula transform and discrete feature extraction, three patient subgroups were identified. The subtype demonstrating high inter-dependency of diffusion and perfusion displayed higher lactate than the other two subtypes (P = 0.016 and P = 0.044, respectively). Both subtypes of low and high inter-dependency showed worse PFS compared to the intermediate subtype (P = 0.046 and P = 0.009, respectively). V. Using a radiomics approach, advanced physiological images showed better performance than structural images for predicting O6-methylguanine-DNA methyltransferase (MGMT) methylation status. For predicting 12-month PFS, the model of radiomic features and clinical factors outperformed the model of MGMT methylation and clinical factors (P = 0.010). In summary, pre-operative multi-parametric MRI shows potential for the non-invasive evaluation of glioblastoma heterogeneity, which could provide crucial information for patient care.

L'objectif de cette thèse est de développer et d'évaluer des paramètres quantitatifs de l'IRM de la prostate en discriminant les cancers de score de Gleason (GS) ≥7. Nous supposons que les paramètres quantitatifs de l'IRM pourraient aider à standardiser le diagnostic, et à diminuer la variation inter-lecteur et/ou inter-institution du diagnostic du cancer de la prostate. Cette thèse est divisée en trois chapitres. Le premier chapitre, intitulé « IRM T2 quantitatif de la prostate », est une étude rétrospective sur une base de données des patients avant prostatectomie radicale. Le deuxième chapitre, intitulé « IRM multiparamétrique quantitative de la prostate », est aussi une étude rétrospective avant prostatectomie radicale. Le troisième chapitre, intitulé « Élastographie IRM de la prostate par voie trans-périnéale» est une étude expérimentale. Notre première étude montre que le T2 est robuste sur les machines de constructeurs différents. Le T2 est un prédicteur significatif, mais de faible performance, d'agressivité du cancer de la prostate à 3T. Notre deuxième étude montre que la combinaison du 10ème centile de l'ADC avec le Time-topeak (TTP) améliore la performance du diagnostic, et ce modèle est lui aussi robuste entre des machines de constructeurs différents. Notre troisième étude montre les résultats préliminaires sur l'élasticité de la prostate. Ces résultats montrent que l'élastographie IRM de la prostate en haute fréquence d'excitation (>100 Hz) par voie trans-périnéale est faisable. L'élastographie pourrait à l'avenir être intégrée à l'IRM multiparamétrique quantitative pour améliorer la performance de diagnostic / The purpose of this thesis is to develop and evaluate the quantitative methods of multiparametric MRI of prostate in discriminating Gleason score (GS) ≥7 cancers. We suppose that the quantitative parameter of MRI could help standardizer the diagnostic, reduce the inter-lecture and/ or inter-institution variation in diagnostic of prostate cancer. This thesis is divided into three chapters. The firs chapter, entilted « Quantitative T2 MRI of prostate » is a retrospective study on a database of prostate cancer patients before radical prostatectomy. The second chapter, entilted « Multi-parametric Quantitative MRI of prostate » is also a retrospective study before radical prostatectomy. The third chapter, entitled « MR elastography of prostate by transperineal approach », is an experimental study. Our first study shows that T2 value is robust between machines of different constructors. T2 value is significant predictor, but of weak performance, of aggressively cancer of prostate at 3T. Our second study shows that the combination of ADC_10th percentile with Time-to-peak (TTP) improved the diagnosis performance, and this model is also robust between two machines of different constructors. Our third study shows the initial results on elasticity of the prostate. These results show that MRI elastography of prostate at high excitation frequency (>100 Hz) by trans-perineale approach was feasible. The elastography may, in the future, be integrated in quantitative multi-parametric MRI to improve the diagnosis performance

Objectives: The study aimed to evaluate the prediction of Prostate Imaging Reporting and Data System (PI-RADS) with respect to the prostate cancer (PCa) detection rate and tumor aggressiveness in magnetic resonance imaging (MRI)/ultrasound-fusion-biopsy (fusPbx) and in systematic biopsy (sysPbx). Materials and Methods: Six hundred and twenty five patients undergoing multiparametric MRI were investigated. MRI findings were classified using PI-RADS v1 or v2. All patients underwent fusPbx combined with sysPbx (comPbx). The lesion with the highest PI-RADS was defined as maximum PI-RADS (maxPI-RADS). Gleason Score ≥ 7 (3 + 4) was defined as significant PCa. Results: The overall PCa detection rate was 51% ( n = 321; 39% significant PCa). The detection rate was 43% in fusPbx ( n = 267; 34% significant PCa) and 36% in sysPbx ( n = 223; 27% significant PCa). Nine percentage of significant PCa were detected by sysPbx alone. A total of 1,162 lesions were investigated. The detection rate of significant PCa in lesions with PI-RADS 2, 3, 4, and 5 were 9% (18/206), 12% (56/450), 27% (98/358), and 61% (90/148) respectively. maxPI-RADS ≥ 4 was the strongest predictor for the detection of significant PCa in comPbx (OR 2.77; 95% CI 1.81–4.24; p < 0.005). Conclusions: maxPI-RADS is the strongest predictor for the detection of significant PCa in comPbx. Due to a high detection rate of additional significant PCa in sysPbx, fusPbx should still be combined with sysPbx.