Parametric mapping model for bladder using free-form deformation

Virtual Endoscopy is a perspective tool for noninvasive tumor detection inside the bladder compared with optical endoscopy. Since the shape, size, position, and relations of a bladder are dependent on the amount of urine, the body position and the age of the person, it would be difficult to compare morphological or intensity features of patient's bladder at different filling stages or bladders with patients for abnormalities detection. In this paper, we present a parametric R3 to R3 mapping model to normalize the bladders with diverse geometric shapes of different size using Free-Form Deformation (FFD). After the establishment of an adaptive FFD to a specified bladder, various image features, such as original morphological and textural attributes of the bladder, are projected onto the model surface and then are further mapped to a Generalized Gaussian sphere. In this way, the proposed method provides a unified way to evaluate the entire bladder wall. The experimental results based on MRI data indicate the feasibility of the bladder mapping using the proposed model. From the surface of the mapped model, the thickness of the bladder can be observed clearly. Moreover, thickened regions which reflect the tumor locations are quantified and detected more easily.