Anatomical Shape and Motion Reconstruction from Sparse Image Data

In current clinical practice, medical imaging plays a key role in diagnosis, therapy planning and therapy monitoring. Some of these modalities, such as CT, MRI, and 3D ultrasound, provide high resolution volumetric anatomical information, and more recently, 3D imaging in time. In certain practical situations, however, limitations with respect to imaging time, space, radiation dose, or ergonomics make it impossible to acquire such rich data. In such cases, imaging may be performed that is of lower dimensionality than the desired information, or is sparse in at least one of the dimensions. This type of sparse imaging is investigated in this thesis. Sparse imaging is typically employed in image guided interventions and surgeries, where high speed, high image resolution and an open acquisition setup are of major importance. The 3D position of the surgical instruments with respect to the 3D patient anatomy is then assessed through 2D imaging such as X-ray fluoroscopy or Ultra-soun. For similar reasons mono- and biplane X-ray fluoroscopy became the standard for kinematic analysis of joints, allowing the acquisition of a wide range of motions, such as running and jumping. In other situations, radiation dose and cost reduction play a mayor role in employing sparse imaging. Bone surface reconstruction from points pin-pointed during knee surgery e.g. has been investigated for replacing prior CT acquisition. Also, assessment of organ motion, such as cardiac or respiratory motion, can occur from temporally sparse CT images.

• View at Worldcat

Free Full Text ( Final Version , 55mb )