X-ray fluoroscopy is the standard imaging modality for guidance of percutaneous catheter-based interventions such as angioplasty and stenting. Its high spatial and temporal resolution allow excellent visualization of interventional devices like guidewires; but it suffers from low contrast of soft tissues. Typical applications require opacification of volumes with radio-contrast agent to distinguish structures; but this is not always possible, as in the case of completely occluded vessels with no blood flow. Also, x-ray fluoroscopy does not allow for characterization of tissues, for example to distinguish healthy from infarcted myocardium. Magnetic resonance imaging (MRI), on the other hand, is characterized by its exceptional and varied modes of soft tissue contrast, though its spatial and temporal resolution are not optimal for guidance of catheter-based procedures. This thesis presents a system of X-ray Fused with MRI (XFM) which aims to exploit the strengths of both modalities to improve the quality of image-guidance and foster development of novel procedures not possible with either modality alone. Registration of the two types of images was based on external fiducial markers attached to the subject's skin. Validation experiments in a vascular phantom showed registration accuracy in the sub-millimeter range. Registration of x-ray and MR images requires correction of distortions present in both images. MRI systems typically have built-in algorithms for distortion correction, but this is not the case for x-ray fluoroscopy. A new method of x-ray distortion correction was developed which allowed for correction of images acquired over a large range of C-arm orientations. It is shown here that only 75 images of a grid phantom were necessary to characterize the x-ray distortion over a range of +/-45° of primary angle, +/-36° of secondary angle, and 98-118 cm of source-to-intensifier distance, with residual errors as low as 0.16 mm.; These XFM methods were implemented in a comprehensive image fusion system and integrated into a commercial x-ray fluoroscopy unit. In vivo validation of targeted endomyocardial injection of therapeutic cells was performed in a swine model of myocardial infarction. Mesenchymal stromal cells were labeled with an iron contrast agent for their visualization on post, injection MRI. This allowed for comparison of the observed location of the injections with their XFM-predicted locations. A total of 130 injections were safely performed in 12 animals with myocardial infarctions 3-74 days old. (Abstract shortened by UMI.)
展开▼