The field of computer-assisted, image-guided surgery is in the formative stages. Neurosurgery has been at the forefront in this technology, and is the field that promises perhaps the greatest challenges and greatest potential clinical benefits. Experience with rigid stereotactic localizing devices and standard 2-D images has shown that conventional surgery can be performed through less invasive approaches, with greater precision, and with shorter operating room and hospitalization times. The preparation of source imaging data for image-guided procedures is a significant impediment to the transfer of this technology from the laboratory to routine clinical practice. The critical bridge between image acquisition and effective image-guided surgery is the use of the computer to recognize and to delineate anatomy. This is the process of segmentation. Current segmentation tools are both manually and computationally intensive, and require substantial user supervision. The goal of this proposal is to develop further a core of robust """"""""smart"""""""" segmentation tools for computer-assisted surgery. Preliminary data from the Surgical Planning Laboratory (SPL) have resulted in increasingly robust, reliable, and facile segmentation algorithms. The applicants proposed to exploit the unique material and human resources of the SPL in order to develop further these segmentation algorithms into stand-alone computer tools that are simple and readily adapted to image-guided neurosurgery. The segmentation algorithms will be enhanced and expanded, and protocols will be developed to assess their reproducibility, accuracy, and precision, based on currently accepted standard comparative measures.
Showing the most recent 10 out of 41 publications
