The size and geometry of the prostate are known to be pivotal quantities usedby clinicians to assess the condition of the gland during prostate cancerscreening. As an alternative to palpation, an increasing number of methods forestimation of the above-mentioned quantities are based on using imagery data ofprostate. The necessity to process large volumes of such data creates a needfor automatic segmentation tools which would allow the estimation to be carriedout with maximum accuracy and efficiency. In particular, the use of transrectalultrasound (TRUS) imaging in prostate cancer screening seems to be becoming astandard clinical practice due to the high benefit-to-cost ratio of thisimaging modality. Unfortunately, the segmentation of TRUS images is stillhampered by relatively low contrast and reduced SNR of the images, therebyrequiring the segmentation algorithms to incorporate prior knowledge about thegeometry of the gland. In this paper, a novel approach to the problem ofsegmenting the TRUS images is described. The proposed approach is based on theconcept of distribution tracking, which provides a unified framework formodeling and fusing image-related and morphological features of the prostate.Moreover, the same framework allows the segmentation to be regularized viausing a new type of "weak" shape priors, which minimally bias the estimationprocedure, while rendering the latter stable and robust.
展开▼
