Color image segmentation for Bladder Cancer Diagnosis

In this paper, a new color image segmentation method is proposed to extract the region of bladder tumor from a color bladder image. In the past, the diagnosis of bladder tumors mainly relies upon cystoscopic examination with an in vivo staining technique. This manner heavily depends on the interpreter's experience and often results in errors in diagnosis. Instead of previous method, we use methylene blue in vivo staining combined with color segmentation techniques to improve the accuracy of the diagnosis of bladder tumors. The segmentation task can be decomposed into two stages. First, cluster detection combined with probabilistic relaxation is used to extract the clusters of specified colors from the HLS color space. Then, in order to refine the chromatic properties, the Bayesian algorithm is employed to reject the false region from the clusters obtained in the first stage. Experimental results show that the proposed method can segment the bladder tumor successfully. The technique could serve as an auxiliary tool for doctors or researchers in the diagnosis of bladder tumors.