Due to highly variable behavior of the prostate cancer, treatment has been hampered by problems inherent in the conventional diagnosis methodology, leading to a serious dilemma: treatment is still too late in some patients, while others are, perhaps, being treated unnecessarily. The detection of prostate cancer has been dramatically increased through improved screening programs. The only completely reliable method for the diagnosis of prostate cancer is said to be through pathological examination of tissue samples. Scientific evidence indicates that one out of five cancers will be missed in needle biopsy, and the accuracy of findings estimated by the existing protocols is unsatisfactory in helping determine the best treatment plan. The general goal of this project is to demonstrate the effectiveness of a statistical master model in prostate biopsy site selection for improved cancer detection and appropriate tumor grade and stage. The main hypothesis to be tested is that statistical modeling and information visualization of localized prostate cancer will better define and evaluate biopsy protocols, leading to an increased cancer detection and an improved accuracy of tumor diagnosis. Because the disease patterns of localized prostate cancer have not previously been defined in a 3D and probabilistic manner, the specific aims of this project are: l) to improve the understanding of the disease patterns through statistical modeling and visualization of localized prostate cancer; 2) to simulate biopsy procedures and correlate the findings with true tumor parameters to determine the accuracy and pitfalls of current techniques; and 3) to recommend new selective biopsy strategies through an optimized needle site selection by correlation of the master model and imaging features.
