Em tecidos biológicos lesados, a viscoelasticidade é a propriedade física que mais se modifica em relação ao tecido normal. Palpação manual é geralmente usada para identificar estas lesões, tais como nódulos e cistos. Recentemente, vários estudos envolvendo técnicas ultrassônicas e de ressonância magnética, denominadas elastografia, têm sido empregadas para avaliar as propriedades viscoelásticas dessas lesões. Uma das dificuldades neste tipo de estudo está relacionada ao desenvolvimento de simuladores de tecidos biológicos com inclusões equivalentes, uma vez que, essas estruturas lesadas originam-se a partir do próprio tecido biológico. Com base nessas motivações, técnicas quantitativas de elastografia por ressonância magnética e ultrassom foram exploradas para avaliar lesões simuladas em fantomas viscoelásticos. Estas lesões, com diferentes propriedades viscoelásticas, foram geradas no interior de um fantoma usando radiação ionizante. Os fantomas, também conhecidos como gel dosimétrico, foram desenvolvidos a base de pele animal, e irradiados utilizando um sistema de terapia de radiação convencional. Imagens de relaxometria por ressonância magnética (RRM) foram adquiridas nestes fantomas e usadas como referência padrão da dose absorvida e de sua distribuição. Os perfis da distribuição de dose avaliados nessas imagens de RRM e pelo sistema de planejamento radioterápico TPS 3D foram comparados aos de rigidez das imagens elastográficas. O estudo elastográfico nestes fantomas foi realizado utilizando os métodos de Vibroacustografia (VA), Vibrometria por Ultrassom (VU) e Elastografia por Ressonância Magnética (ERM). O segundo objetivo desta tese foi explorar a viabilidade de existência de speckle em imagens de vibroacustografia. Para testar essa idéia, um modelo tridimensional (3D) para a função de espalhamento de ponto (PSF) do sistema de VA foi simulada. O código da simulação foi desenvolvido em ambiente MATLAB e empregando sub-rotinas do programa Field II para simulação numérica dos transdutores. Imagens de ultrassom modo-B (IUSB) e vibroacustografia foram simuladas usando esse modelo de PSF-3D para um transdutor esférico e confocal, respectivamente. Essas IUSB foram simuladas para servir como um parâmetro comparativo com as imagens geradas por VA. As imagens de ultrassom foram exibidas em um plano tomográfico que corresponde ao plano de imagem da VA. As simulações foram realizadas utilizando um cluster de computadores de alto desempenho. Todas as imagens foram simuladas empregando um modelo de fantoma virtual não homogêneo com dimensões de (10 × 10 × 50) mm3. Os resultados preliminares mostraram um padrão de interferência nas imagens de VA, semelhantes à speckles, obtidas empregando o transdutor confocal. Estas imagens foram produzidas a partir de espalhadores localizados no interior do volume da célula de resolução da PSF-3D. / The viscoelasticity of injured biological tissues is the physical property that changes the most in relation to normal tissue. Manual palpation is commonly used to identify these lesions, such as nodules and cysts. Recently, several studies involving ultrasound and magnetic resonance imaging techniques, called elastography, have been employed to assess the viscoelastic properties of these lesions. One difficulty in this type of study is related to the development of biological tissues mimicking materials with similar inclusions, once these injured structures originate from anomalies within the biological tissue. Based on these motivations, quantitative elastographic techniques based on magnetic resonance and ultrasound modalities have been used to assess injuries in viscoelastic mimicking-tissue phantom materials. These lesions, with different viscoelastic properties, were generated within a phantom using ionizing radiation. The phantoms, also known as dosimeter gel, were developed based on animal skin powder, and irradiated using a conventional radiation therapy system. Magnetic resonance relaxometry images (MRR) were acquired in these phantoms, and were used as absorbed dose standard reference and its distribution. The estimated dose distribution profiles from these images and the ones provided by TPS 3D software radiotherapy planning system were compared to the elastograms. The elastographic studies were conducted using Vibro-acoustography (VA), Magnetic Resonance Elastography (MRE) and Shearwave Dispersion Ultrasound Vibrometry (SDUV) techniques. The second goal of this thesis was to explore the feasibility of speckle existence in vibro-acoustography images. To test that, a three-dimensional (3D) model for the systems point spread function (PSF) was simulated. The simulation code was implemented in MATLAB and using the program Field II subroutines. The numerical simulations were performed using a cluster of high performance computers. B-mode ultrasound and VA images were simulated using that PSF 3D model, for a spherically focused and a confocal transducer, respectively. These B-mode images were simulated as a comparative parameter to the images generated by VA. The B-mode ultrasound images were displayed in a tomographic plane corresponding to the VA imaging plane. All images were simulated using a virtual phantom with dimensions (10 × 10 × 50) mm3. Preliminary results showed a interference pattern in VA images taken with a confocal transducer. These images were produced from scatterers located inside the PSF resolution cell volume.

Cette thèse a été réalisée dans le cadre de la dosimétrie des faisceaux de haute énergie délivrés au patient pendant un traitement de radiothérapie externe. L'objectif de ce travail est de vérifier que la distribution de dose 3D absorbée dans le patient est conforme au calcul réalisé sur le système de planification de traitement (TPS) à partir de l'imageur portal (en anglais : Electronic Portal Imaging Device, EPID). L'acquisition est réalisée en mode continu avec le détecteur aS-1200 au silicium amorphe embarqué sur la machine TrueBeam STx (VARIAN Medical system, Palo Alto, USA). Les faisceaux ont une énergie de 10 MeV et un débit de 600 UM.min-1. La distance source-détecteur (DSD) est de 150 cm. Après correction des pixels défectueux, une étape d'étalonnage permet de convertir leur signal en dose absorbée dans l'eau via une fonction de réponse. Des kernels de correction sont également utilisés pour prendre en compte la différence de matériaux entre l'EPID et l'eau et pour corriger la pénombre sur les profils de dose. Un premier modèle de calcul a permis ensuite de rétroprojeter la dose portale en milieu homogène en prenant en compte plusieurs phénomènes : les photons diffusés provenant du fantôme et rajoutant un excès de signal sur les images, l'atténuation des faisceaux, la diffusion dans le fantôme, l'effet de build-up et l'effet de durcissement du faisceau avec la profondeur. La dose reconstruite est comparée à celle calculée par le TPS avec une analyse gamma globale (3% du maximum de dose et 3 mm de DTA). L'algorithme a été testé sur un fantôme cylindrique homogène et sur un fantôme de pelvis à partir de champs modulés en intensité (RCMI) et à partir de champs d'arcthérapie volumique modulés, VMAT selon l'acronyme anglais Volumetric Modulated Arc Therapy. Le modèle a ensuite été affiné pour prendre en compte les hétérogénéités traversées dans le milieu au moyen des distances équivalentes eau dans une nouvelle approche de dosimétrie plus connue sous le terme de " in aqua vivo " (1). Il a été testé sur un fantôme thorax et, in vivo sur 10 patients traités pour une tumeur de la prostate à partir de champs VMAT. Pour finir, le modèle in aqua a été testé sur le fantôme thorax avant et après y avoir appliqué certaines modifications afin d'évaluer la possibilité de détection de sources d'erreurs pouvant influencer la bonne délivrance de la dose au patient.[...] / This thesis aims at the dosimetry of high energy photon beams delivered to the patient during an external radiation therapy treatment. The objective of this work is to use EPID the Electronic Portal Imaging Device (EPID) in order to verify that the 3D absorbed dose distribution in the patient is consistent with the calculation performed on the Treatment Planning System (TPS). The acquisition is carried out in continuous mode with the aS-1200 amorphous silicon detector embedded on the TrueBeam STx machine (VARIAN Medical system, Palo Alto, USA) for 10MV photons with a 600 UM.min-1 dose rate. The source-detector distance (SDD) is 150 cm. After correction of the defective pixels, a calibration step is performed to convert the signal into an absorbed dose in water via a response function. Correction kernels are also used to take into account the difference in materials between EPID and water and to correct penumbra. A first model of backprojection was performed to reconstruct the absorbed dose distribution in a homogeneous medium by taking into account several phenomena: the scattered photons coming from the phantom to the EPID, the attenuation of the beams, the diffusion into the phantom, the build-up, and the effect of beam hardening with depth. The reconstructed dose is compared to the one calculated by the TPS with global gamma analysis (3% as the maximum dose difference criteria and 3mm as the distance to agreement criteria). The algorithm was tested on a homogeneous cylindrical phantom and a pelvis phantom for Intensity-Modulated Radiation Therapy (IMRT) and (Volumetric Arc Therapy (VMAT) technics. The model was then refined to take into account the heterogeneities in the medium by using radiological distances in a new dosimetrical approach better known as "in aqua vivo" (1). It has been tested on a thorax phantom and, in vivo on 10 patients treated for a prostate tumor from VMAT fields. Finally, the in aqua model was tested on the thorax phantom before and after making some modifications to evaluate the possibility of detecting errors that could affect the correct delivery of the dose to the patient. [...]

科学研究費補助金 研究種目:基盤研究(C)(2) 課題番号:07680528 研究代表者:青山 隆彦 研究期間:1995-1996年度