Multi-circle detection for bladder cancer diagnosis based on artificial immune systems

Bladder cancer is the fourth most common type of cancer in men and the ninth in women in United States. A recent approach for early bladder cancer detection is to mix human urine samples with some very small beads that are coated with special biochemical materials which can bind to tumor cells, but not to normal cells. By examining and analyzing bead images of urine samples under a microscope, patients with potential cancer risk can be identified. Multi-circle detection is a challenging problem for processing bead images in an automatic bladder cancer diagnosis system, due to the large number and non-ideal shapes of objects (e.g., beads with cancer cells) in microscope images. In this study, a new approach based on real valued artificial immune system is developed and tested. Computer simulation results show that this algorithm outperforms traditional methods such as circular Hough Transform and geometric characteristic based methods in terms of both precision and robustness.