Parametric Shape Modeling Using Deformable Superellipses for Prostate Segmentation

Abstract

Automatic prostate segmentation in ultrasound images is a challenging task due to speckle noise, missing boundary segments, and complex prostate anatomy. One popular approach has been the use of deformable models. For such techniques, prior knowledge of the prostate shape plays an important role in automating model initialization and constraining model evolution. In this paper, we have modeled the prostate shape using deformable superellipses. This model was fitted to 594 manual prostate contours outlined by five experts. We found that the superellipse with simple parametric deformations can efficiently model the prostate shape with the Hausdorff distance error (model versus manual outline) of 1.32 +/- 0.62 mm and mean absolute distance error of 0.54 +/- 0.20 mm. The variability between the manual outlinings and their corresponding fitted deformable superellipses was significantly less than the variability between human experts with p-value being less than 0.0001. Based on this deformable superellipse model, we have developed an efficient and robust Bayesian segmentation algorithm. This algorithm was applied to 125 prostate ultrasound images collected from 16 patients. The mean error between the computer-generated boundaries and the manual outlinings was 1.36 +/- 0.58 mm, which is significantly less than the manual interobserver distances. The algorithm was also shown to be fairly insensitive to the choice of the initial curve.

3D Prostate Segmentation in Ultrasound Images Based on Tapered and Deformed Ellipsoids

“…Deformable models have been widely used for medical image segmentation [5,6,7,8,9] and are generally more successful than the former methods. Fitting ellipses, ellipsoids, super-ellipses, and deformable ellipses or using them for initialization have been relatively attractive approaches for prostate segmentation due to the shape of the gland [10,11,12,13,14].…”

“…While the extensive study of [12] showed good segmentation results, the method presented is limited to 2D. Previous 3D methods [15,16] are time consuming (> 2 minutes) or require significant user intervention [11].…”

Prostate segmentation with local binary patterns guided active appearance models

“…The model selectively discarded salient points for building the shape model to improve segmentation accuracy in presence of shadow artifacts in TRUS images. Gong et al 8 proposed to use deformable super ellipse to segment the prostate. Shape constraints of the model was effective in achieving impressive segmentation results.…”

Segmentation of elastographic images using a coarse-to-fine active contour model

“…Although there is little literature on automated segmentation of treated sites during RF ablation, segmentation techniques for US images have been reported for applications with echocardiographic (e.g., Corsi et al 2002;Angelini et al 2005), breast (e.g., Horsch et al 2002;Chang et al 2003a) and prostate data (e.g., Shen et al 2003;Gong et al 2004). Because of characteristic US artefacts, such as speckle and shadowing, intensity inhomogeneities, low contrast and ill-defined boundaries, simple image feature-based thresholding or edge-detection methods are ineffective.…”

“…Because of characteristic US artefacts, such as speckle and shadowing, intensity inhomogeneities, low contrast and ill-defined boundaries, simple image feature-based thresholding or edge-detection methods are ineffective. Successful segmentation algorithms reported for US images are based on morphologic operations (Czerwinski et al 1999;Gong et al 2004), neural networks (Binder et al 1999), wavelet analysis (Angelini et al 2001) and Markov random fields (Haas et al 2000;Xiao et al 2002;Brusseau et al 2004;Gong et al 2004). These incorporate preprocessing for speckle reduction (e.g., the "stick" method) (Czerwinski et al 1999), anisotropic diffusion (Perona and Malik 1990) and intensity corrections (Xiao et al 2002).…”