Image processing tools with applications to prostate cancer diagnosis and Gleason grading
Prostate cancer remains the most common frequently diagnosed cancer and the second leading cause of cancer death in adult men. In the US, recent cancer estimates for 2010 indicated an incidence rate of 157 case in each 100,0000 men with approximately 218,000 new diagnosed cases and 32,000 deaths each year. Currently, pathologists visually grade prostate biopsy tissues using Gleason scoring system. This approach is very subjective in nature and subject to observer's variations. Also, it is a time consuming task and raises difficulties as far as spatial resolution is considered especially in the subgroups of grades 3--5 where further classification is needed. In addition, pathologists experience is another factor that contributes to errors in reporting accurate Gleason grades. In this research, image processing tools with applications to prostate cancer diagnosis and Gleason grading are presented. The research is aimed at providing a computer-aided classification system for prostate cancer pathological images through the use of image processing and feature extraction algorithms. The Classification is based on Gleason grading system where the highly common prostate cancer Gleason grades 3--5 are considered. The developed classification system achieves a higher recognition rate compared to existing systems in literature and provides a new prostate biopsy visualization tool.