Permanent implantation of radioactive seeds is a viable and effective therapeutic option widely used today for early-stage prostate cancer. Compared to external radiation therapy in which radiation must penetrate healthy tissues in order to reach cancer cells, implantation of low-energy radionuclides permits highly localized delivery of radiation. Although the implant procedure has improved in recent years with the help of computerized treatment planning and image guidance techniques, significant enhancement of clinical outcome is expected from implementation of real-time intraoperative dosimetry and optimization. Intraoperative evaluation of dose delivery would permit identification of underdosed regions and remedial seed placement, thus ensuring that the entire prostate volume receive the prescribed dose. However, before the concept can be realized, the problem of real-time seed localization must be solved. This is the focus of this investigation.
The specific aims of the project include development of (1) a fully automated method to segment seed images from the fluoroscopic data and (2) a fully automated method to identify individual seed positions including those that are superposed. The image segmentation will be accomplished by a region based adaptive thresholding technique. Subsequent localization of the seeds will be performed by a hierarchical decision process aided by an artificial intelligence controlled seed classifier.
Lam, Steve T; Cho, Paul S; Marks 2nd, Robert J et al. (2004) Three-dimensional seed reconstruction for prostate brachytherapy using Hough trajectories. Phys Med Biol 49:557-69 |
Narayanan, Sreeram; Cho, Paul S; Marks 2nd, Robert J (2004) Three-dimensional seed reconstruction from an incomplete data set for prostate brachytherapy. Phys Med Biol 49:3483-94 |