Le traitement des arythmies cardiaques par ablation radiofréquence est une procédure thérapeutique permettant de restaurer un rythme normal par destruction thermique des tissus arythmogènes. A l'heure actuelle, l'intervention est réalisée sans imagerie temps réel permettant de visualiser la lésion pendant l'ablation. La thermométrie IRM permet de mesurer la température du tissu en chaque pixel et d'estimer directement l'étendue de la lésion via le calcul de la dose thermique cumulée. Si cette technique est déjà établie pour guider l'ablation de tumeurs dans différents organes, elle reste difficile à mettre en œuvre sur le cœur, notamment à cause des mouvements de respiration et de contraction myocardique. Dans le cadre de cette thèse, une méthode de thermométrie cardiaque a été implémentée pour réaliser une cartographie temps réel de la température du myocarde en condition de respiration libre. Plusieurs séquences IRM rapides ont été développées pour permettre l'acquisition d'environ 5 coupes par battement cardiaque avec une taille de voxel de 1.6X1.6X3 mm3. Plusieurs solutions de réduction des mouvements hors plan de coupe ont été évaluées : positionnement des coupes dans le sens principal du déplacement, suivi dynamique de la position des coupes en fonction de l'état respiratoire (navigateur, mesure de la position du cathéter). Le mouvement résiduel et les artéfacts de susceptibilité associés sont corrigés par des algorithmes temps réels pour permettre une précision de la thermométrie IRM à ±2°C sur les ventricules. Ce protocole a été utilisé avec succès pour le suivi d'ablations radiofréquences chez la brebis (N=3), permettant une corrélation (R=0.87) entre la dose thermique et la taille réelle des lésions induites. Les résultats sont très prometteurs quant à la pertinence de cette mesure pour une estimation en ligne de l'étendue de la lésion pendant l'ablation. Ces méthodes permettent d'envisager une évaluation clinique à court terme. / Radiofrequency ablation is a therapeutic procedure for the treatment of cardiac arrhythmia by inducing a local necrosis of the arrhythmogenic tissue. This intervention is currently performed without online imaging of the lesion formation during radiofrequency delivery. MR thermometry provides a monitoring of temperature in the targeted tissue in each pixel and an immediate estimation of lesion via the calculation of the thermal dose. If this technique is well established for the guidance of tumor ablation in various organs, it remains challenging in the heart due to motion (breathing and myocardial contraction). In this work, a cardiac MR thermometry method was developed to perform a real-time temperature mapping of the heart under free-breathing conditions. Several MR pulse sequences were designed to ensure the acquisition of up to 5 slices per heartbeat with a voxel size of 1.6X1.6X3 mm3. Different solutions of minimization of out-of-plane motion were evaluated: alignment of the slices in the main direction of displacement, dynamic update of slice position depending on the respiratory state (echo-navigator, measure of the catheter position). Residual in-plane motion and associated susceptibility artifacts were corrected by real-time algorithms to allow a precision of MR thermometry of ±2°C in ventricles. This protocol was successfully used for the monitoring of radiofrequency ablation in sheep (N=3), allowing a correlation (R=87) between thermal dose maps and sizes of created lesions. These results are promising regarding the relevance of this measure for an inline estimation of the lesion extent during ablation.

No description available.

Les travaux présentés dans ce manuscrit s’inscrivent dans un projet de thermométrie cardiaque par IRM. L’objectif de cette approche est de monitorer en temps réel le traitement des arythmies cardiaques par ablation thermique. Dans la chaine de traitement utilisée, des méthodes de recalage d’images doivent être implémentées afin d’assurer un suivi exploitable de l’ablation. En effet, les mouvements des différents organes localisés dans la (les) coupe(s) d’imagerie ont un impact sur la précision de la thermométrie IRM. Les méthodes d’estimation de mouvement doivent cependant être robustes aux différents artefacts (faible SNR, flux sanguin dans les cavités cardiaques, etc.) et utilisables en pseudo temps réel (10 images/seconde). Les travaux présentés dans ce manuscrit se concentrent sur la robustesse des méthodes de recalage d’images. Tout d’abord, le bruit inhérent aux images IRM a été intégré à l’estimation de mouvement. Ceci permet de pondérer localement les voxels de l’image dans le calcul du déplacement. Ensuite, une annulation numérique du signal chaotique du sang à l’intérieur du ventricule gauche est proposée via une segmentation semi-automatique de celuici à base de modèle déformable. Un nouveau terme, issu de la probabilité d’appartenir à ce ventricule, a été ajouté dans l’algorithme de contour actif. Les méthodes proposées apportent une amélioration de la qualité de l’estimation du mouvement. Elles sont adaptées à la chaine de traitement de thermométrie afin de les rendre automatiques dans la phase d’ablation thermique de la procédure. De plus, ces méthodes répondent aux contraintes de temps réel de la thermométrie IRM / The works presented in the manuscript are incorporated within the framework of cardiac MR thermometry. The aim of this approach is to monitor in real time the treatment of arrhy²thmias by thermal ablation. In the pipeline used in thermometry, image registration methods have to be inserted to unsure a reliable monitoring of the treatment. Indeed, motions of the different organs present in the acquisition slices have an impact on the accuracy of the MR thermometry. Furthermore, image registration has to be robust to artefacts (low SNR, blood flow in the heart cavities…) and has to be used in real time (10 images/second). The works presented in this manuscript focuses on the robustness of the image registration methods. First, the MR images inherent noise is integrated to the motion estimation. It enables the local weighting of the image’s voxels in the computing of the displacement. Then, a numerical cancelation of the chaotic blood flow signal within the left ventricle through a semi-automatic segmentation is proposed. A new term based on the probability to belong to this ventricle is added to the active contour algorithm. The proposed methods improved the quality of the motion estimation. They are adapted to the thermometry pipeline to make them automatic in the thermal ablation phase of the procedure. They are also compatible with the real-time aspect of MR thermometry.

The clinical efficacy of chemotherapy in solid tumours is limited by systemic toxicity and the inability to deliver a cytotoxic concentration of anticancer drugs to all tumour cells. Temperature sensitive drug carriers provide a mechanism for triggering the rapid release of chemotherapeutic agents in a targeted region. Thermally mediated drug release also leverages the ability of hyperthermia to increase tumour blood flow, vessel permeability, and drug cytotoxicity. Drug release from thermosensitive liposome drug carriers in the tumour vasculature serves as a continuous intravascular infusion of free drug originating at the tumour site. However, localized drug release requires precise heating to improve drug delivery and efficacy in tumours while minimizing drug exposure in normal tissue. Focused ultrasound can noninvasively heat millimeter-sized regions deep within the body, and can be combined with MR thermometry for precise temperature control. This thesis describes the development of strategies to achieve localized hyperthermia using MRI-controlled focused ultrasound, for the purpose of image-guided heat-triggered drug release from thermosensitive drug carriers. First, a preclinical MRI-controlled focused ultrasound system was developed as a platform for studies of controlled hyperthermia and drug delivery in rabbits. The feasibility of using ultrasound hyperthermia to achieve localized doxorubicin release from thermosensitive liposomes was demonstrated in normal rabbit muscle. Second, strategies were described for using MR thermometry to control ultrasound heating at a muscle-bone interface based on MR temperature measurements in adjacent soft tissue, demonstrating localized drug delivery in adjacent muscle and bone marrow. Third, fluorescence microscopy was employed to demonstrate that increased overall drug accumulation in rabbit VX2 tumours corresponds to high levels of bioavailable drug reaching their active site in the nuclei of tumour cells. The results of this thesis demonstrate that image-guided drug delivery using thermosensitive liposomes and MRI-controlled focused ultrasound hyperthermia can be used to noninvasively achieve precisely localized drug deposition in soft tissue, at bone interfaces, and in solid tumours. Clinical application of this work could provide a noninvasive means of enhancing chemotherapy in a variety of solid tumours.

The clinical efficacy of chemotherapy in solid tumours is limited by systemic toxicity and the inability to deliver a cytotoxic concentration of anticancer drugs to all tumour cells. Temperature sensitive drug carriers provide a mechanism for triggering the rapid release of chemotherapeutic agents in a targeted region. Thermally mediated drug release also leverages the ability of hyperthermia to increase tumour blood flow, vessel permeability, and drug cytotoxicity. Drug release from thermosensitive liposome drug carriers in the tumour vasculature serves as a continuous intravascular infusion of free drug originating at the tumour site. However, localized drug release requires precise heating to improve drug delivery and efficacy in tumours while minimizing drug exposure in normal tissue. Focused ultrasound can noninvasively heat millimeter-sized regions deep within the body, and can be combined with MR thermometry for precise temperature control. This thesis describes the development of strategies to achieve localized hyperthermia using MRI-controlled focused ultrasound, for the purpose of image-guided heat-triggered drug release from thermosensitive drug carriers. First, a preclinical MRI-controlled focused ultrasound system was developed as a platform for studies of controlled hyperthermia and drug delivery in rabbits. The feasibility of using ultrasound hyperthermia to achieve localized doxorubicin release from thermosensitive liposomes was demonstrated in normal rabbit muscle. Second, strategies were described for using MR thermometry to control ultrasound heating at a muscle-bone interface based on MR temperature measurements in adjacent soft tissue, demonstrating localized drug delivery in adjacent muscle and bone marrow. Third, fluorescence microscopy was employed to demonstrate that increased overall drug accumulation in rabbit VX2 tumours corresponds to high levels of bioavailable drug reaching their active site in the nuclei of tumour cells. The results of this thesis demonstrate that image-guided drug delivery using thermosensitive liposomes and MRI-controlled focused ultrasound hyperthermia can be used to noninvasively achieve precisely localized drug deposition in soft tissue, at bone interfaces, and in solid tumours. Clinical application of this work could provide a noninvasive means of enhancing chemotherapy in a variety of solid tumours.

The main purpose of this research was to develop improved methods for RF coil characterisation, and for non-invasive spatio-temporal mapping of temperature in the living body, in order to utilise the full potential of magnetic resonance imaging (MRI) at high magnetic fields by ensuring radiofrequency (RF) safety. Current RF power limits are often overly conservative, unnecessarily limiting the full potential of MRI, especially at high field. Thus it is useful to monitor tissue temperature while running MR imaging sequences which may deposit high RF power. Proton resonance frequency (PRF) MR thermometry can employ the phase of the complex MR signal to estimate temperature change over time. However, the shift of the water PRF with temperature is relatively small, making phase-based MR thermometry inherently sensitive to any extraneously caused changes of local frequency or MR phase. A potential source of error to PRF MR thermometry is a change in surround air susceptibility. The considerable impact of air susceptibility changes on PRF MR thermometry was demonstrated and quantified in experiments and magnetic field simulations. One way of correcting MR thermometry is to use a chemically shifted reference substance, in combination with a phase-sensitive chemical shift-selective MR thermometry sequence. The requirement of having a reliable separation of substances based on their resonance frequency was met by a novel frequency-selective phase-sensitive spin-echo (SE) MR thermometry sequence. This sequence was thoroughly tested in phantom and in-vivo experiments as well as in extensive Bloch simulations. The sequence limitations and advantages are discussed in detail. This technique acquires unsaturated water and fat images in rapid succession at the same position in space. The acquisition of a water and fat slice in less than 100 ms allows the correction of rapid field fluctuations in the brain caused by breathing and heartbeat, while still ensuring the correction of long term drift. With no assumptions required regarding temperature distribution in the tissue, this novel MR thermometry technique can measure brain temperature within a single (1.5 mm)3 voxel with a very low standard deviation (SD) of 0.3 K. Using an MRI phantom with a dimethyl sulfoxide reference, heating experiments achieved a MR temperature measurement with an SD of approximately 0.1 K in a single (1.5 mm)3 voxel. In conclusion, the work presented in this thesis assists the development of a real-time in-vivo temperature monitoring system that guarantees patient RF safety at high field.

Magnetic Resonance Imaging (MRI) is an important diagnostic tool for imaging soft tissue without the use of ionizing radiation. Moreover, through advanced signal processing, MRI can provide more than just anatomical information, such as estimates of tissue-specific physical properties. Signal processing lies at the very core of the MRI process, which involves input design, information encoding, image reconstruction, and advanced filtering. Based on signal modeling and estimation, it is possible to further improve the images, reduce artifacts, mitigate noise, and obtain quantitative tissue information. In quantitative MRI, different physical quantities are estimated from a set of collected images. The optimization problems solved are typically nonlinear, and require intelligent and application-specific algorithms to avoid suboptimal local minima. This thesis presents several methods for efficiently solving different parameter estimation problems in MRI, such as multi-component T2 relaxometry, temporal phase correction of complex-valued data, and minimizing banding artifacts due to field inhomogeneity. The performance of the proposed algorithms is evaluated using both simulation and in-vivo data. The results show improvements over previous approaches, while maintaining a relatively low computational complexity. Using new and improved estimation methods enables better tissue characterization and diagnosis. Furthermore, a sequence design problem is treated, where the radio-frequency excitation is optimized to minimize image artifacts when using amplifiers of limited quality. In turn, obtaining higher fidelity images enables improved diagnosis, and can increase the estimation accuracy in quantitative MRI.

L’hyperthermie guidée par IRM permet l’ablation thermique des tumeurs, l’activation de l’expression d’un transgène sous contrôle d’un promoteur thermo-sensible ainsi que le dépôt local de médicaments à l’aide de nanovéhicules sensibles à la température ou à la pression locale. L’imagerie de température par IRM, basée sur la technique du décalage de la fréquence de résonance du proton permet le monitorage des interventions d’hyperthermie. Les procèdes interventionnels guides par IRM sur cible mobile requièrent des séquences d’imagerie rapides afin d’obtenir des images de phases ayant une résolution spatio-temporelle élevée. Nous avons démontré l’efficacité de l’association des méthodes adaptatives d’imagerie parallèle telles que TSENSE et TGRAPPA et de la méthode multi-référence de l’atlas de mouvement afin de compenser les variations du champ magnétique induites par les organes en mouvement. Les procédés interventionnels guides par IRM sont basés sur des séquences d’imagerie rapides capables de fournir des images en temps-réel ayant une relation précise entre la position de la cible représentée dans l’image et sa vraie position spatiale. Les séquences écho-planar sont très rapides mais possèdent des distorsions géométriques. Nous avons proposé une méthode de correction des distorsions des images EPI. Cette technique est basée sur des approches existantes utilisant l’acquisition de deux images EPI ayant deux temps d’écho différents. L’efficacité de la méthode proposée a été démontrée pour une expérience de thermométrie par IRM. La rapidité du traitement des données, associée à une faible diminution de la rapidité d’acquisition, rend cette méthode particulièrement adaptée pour les procédés interventionnels guides par IRM. La perfusion sanguine, la diffusion thermique ainsi que le coefficient d’absorption des ondes acoustiques ou électromagnétiques déterminent la distribution de la température durant les procédés interventionnels. Certaines tumeurs ont des taux de perfusion élevés conduisant à une évacuation importante de la chaleur et par conséquent, un refroidissement rapide de la cible. Cet effet réduit la température maximale atteinte pour une puissance donne et peut conduire à des zones d’ablation plus petites réduisant ainsi l’efficacité de l’intervention. La connaissance précise des paramètres thermiques du tissu peut aider à la planification des procédés interventionnels. Dans ce but, nous avons proposé une méthode permettant la détermination précise des paramètres cités précédemment. / MR-guided HIFU-induced hyperthermia allows for thermal ablation of tumors, for gene therapy by thermal induction of transgenic expression (based on a thermo-sensitive promoter) and for local drug delivery using thermo-sensitive liposomes. These applications require accurate temperature measurement during the therapeutic intervention. Dynamic MR-temperature imaging based on the proton resonance frequency shift technique allows monitoring the local temperature evolution during hyperthermia. MR-guided thermotherapy on moving organs requires imaging sequences providing phase images with high temporal and spatial resolution. We demonstrated the feasibility of combining adaptive parallel imaging techniques such as TSENSE or TGRAPPA with the atlas-based multi-baseline method for compensating the magnetic field variations produced by moving organs during the respiratory cycle. Many MR-guided interventional procedures rely on real-time imaging sequences for providing precise relations between the target position in the image and the true position in the scanner. Although echo-planar imaging (EPI) sequences are very fast, they are prone to geometric distortions. For correcting these distortions, we proposed a real-time correction method by applying existing approaches based on a dual EPI acquisition with varying echo times. It is demonstrated that this method works well in combination with MR-thermometry for guiding thermal therapies. Short data-processing times as well as a small penalty in acquisition speed make this method well-adapted for MR-guided interventions. Local blood perfusion, thermal conductivity and the absorption coefficient of acoustic or electro-magnetic waves determine the temperature distribution in living tissue. Some tumors have high perfusion rates resulting in considerable heat evacuation. This effect reduces the maximal temperature increase achievable for a given deposited energy and produces smaller ablation zones, which can impair the efficiency of the therapeutic procedure. A method for accurately estimating the above mentioned tissue parameters, was presented. This method could thus be useful in quantifying the influence of perfusion during thermal interventions.