Computers are beginning to enter the realm of surgery, with advances in medical imaging, robotics, and instrumentation leading a trend toward minimally invasive surgical procedures. One of the most prominent examples of the penetration of technology into surgery is radiosurgery. Radiosurgery is a minimally invasive surgical procedure that uses focused beams of radiation to destroy tumors. Though proven effective, its use has been restricted to the treatment of small tumors within the brain. This limitation has little to do with radiosurgery's theoretical applicability to the treatment of a wider variety of tumors, but results instead from the inflexibility of conventional radiosurgical technology. Driven by a desire to extend radiosurgery to treat extracranial tumors and large tumors, a new generation of radiosurgical devices is beginning to emerge. Because these systems provide more flexibility than their predecessors, they increase the complexity of the treatment-planning problem.; I have developed an automatic treatment-planning system, scCARABEAMER, for use with the Cyberknife, a new device for image-guided robotic radiosurgery. scCARABEAMER takes as input a geometric description of the structures of interest, and a set of constraints on the dose to each of these structures; it returns a set of beam configurations that will concentrate a lethal dose of radiation within the tumor, while maintaining the dose to the surrounding structures of interest below the specified level. In planning, it considers the geometric relationships among the tumor, the critical structures, and the treatment beams, and reduces the treatment-planning problem to a linear optimization problem. Its treatment-planning algorithm is independent of the tumor shape, size and location relative to critical structures, and thus is able to generate highly conformal treatment plans for a wide variety of treatment situations.; In this dissertation, I describe the treatment-planning algorithm at the core of scCARABEAMER, and I empirically demonstrate scCARABEAMER's general applicability to a wide range of tumors. I have evaluated scCARABEAMER with respect to two hypotheses: (1) scCARABEAMER can generate treatment plans that meet the current requirements of radiosurgery, and (2) scCARABEAMER is general enough to generate plans for difficult treatment planning situations. The results of my evaluation put scCARABEAMER in a strong position to define the new generation of treatment-planning algorithms that will emerge, in conjunction with new radiosurgical devices, to extend radiosurgery's applicability.
展开▼
