Statistical Shape and Probability Prior Model for Automatic Prostate Segmentation

Abstract

Abstract-Accurate prostate segmentation in Trans RectalUltra Sound (TRUS) images is an important step in different clinical applications. However, the development of computer aided automatic prostate segmentation in TRUS images is a challenging task due to low contrast, heterogeneous intensity distribution inside the prostate region, imaging artifacts like shadow, and speckle. Significant variations in prostate shape, size and contrast between the datasets pose further challenges to achieve an accurate segmentation. In this paper we propose to use graph cuts in a Bayesian framework for automatic initialization and propagate multiple mean parametric models derived from principal component analysis of shape and posterior probability information of the prostate region to segment the prostate. The proposed framework achieves a mean Dice similarity coefficient value of 0.974±0.006, mean mean absolute distance value of 0.49±0.20 mm and mean Hausdorff distance of 1.24±0.56 mm when validated with 23 datasets in a leave-one-patient-out validation framework.

A Fast, Fully Automated Prostate Boundary Segmentation Using Probabilistic Approaches In Ultrasound Images

“…In 2011, Ghose et al [33] presented a new statistical shape model approach using graph cuts in a Bayesian framework for automatic prostate outlining. K-means clustering was used to However, in their methodology, posterior probabilities obtained from MRF modeling were replaced with intensity.…”

“…A quantitative comparison of performance evaluation between the reviewed algorithms (given in Chapter 2) and our proposed method is listed in Shen et al [23] 2003 N/A 500 MHz 64 sec Abolmaesumi [24] 2004 512x512 N/A N/A Gong et al [25] 2004 256x256 P4 -2GHz 5 sec Sahba et al [26] 2005 N/A N/A N/A Zaim [27] 2005 489x382 789MHz 12 sec Badiei et al [28] 2006 480x640 P4 -1.7 GHZ 25.35 sec Zaim et al [29] 2007 489x382 N/A N/A Cosio [30] 2008 512x512 2 GHz 14 min Yan et al [31] 2010 640x480 Core 2 1.86GHz 0.3 sec (C++) Ghose et al [33] 2011 Markov Random Field (MRF) theory is utilized in order to strengthen the boundary edges and to remove the micro-calcifications inside the prostate gland. The probability distribution parameters are obtained from Expectation Maximization algorithm.…”

A combinatorial Bayesian and Dirichlet model for prostate MR image segmentation using probabilistic image features

“…Further, graph-based algorithms have been added to the current methodologies. Ghose et al suggested a scheme that incorporates the graph-cut (GC) algorithm into a Bayesian framework for automated initialization and parametric models (Ghose et al 2011). This scheme, however, involves a number of intermediate steps, which reduce the robustness of the segmentation and the ability to optimize the parameters.…”