Image-asssited system for the diagnosis of bladder tumor recurrence

This paper presents a system for detecting mor-phometric properties of bladder mucosa images obtained with a cystoscope. These properties are calculated for obtaining information about the probability of tumour recurrence. For this purpose, illumination problems of these images have been corrected to apply a vessels segmentation algorithm. Finally, the number of bifurcations of the vascular tree and a ratio Vessels/Bladder area are calculated to validate the hypothesis mentioned above. This validation is carried out with 20 images of bladder mucosa obtaining very promising results, in particular the presented algorithm achieves a sensibility of 0.8571 and a specificity of 1. Consequently, the system proposed in this paper shows valuable properties for the specific objective pointed, i.e, the characterization of the bladder tumour recurrence.