A magnetic resonance spectroscopy driven initialization scheme for active shape model based prostate segmentation

Abstract

Segmentation of the prostate boundary on clinical images is useful in a large number of applications including calculation of prostate volume pre- and post-treatment, to detect extra-capsular spread, and for creating patient-specific anatomical models. Manual segmentation of the prostate boundary is, however, time consuming and subject to inter- and intra-reader variability. T2-weighted (T2-w) magnetic resonance (MR) structural imaging (MRI) and MR spectroscopy (MRS) have recently emerged as promising modalities for detection of prostate cancer in vivo. MRS data consists of spectral signals measuring relative metabolic concentrations, and the metavoxels near the prostate have distinct spectral signals from metavoxels outside the prostate. Active Shape Models (ASM’s) have become very popular segmentation methods for biomedical imagery. However, ASMs require careful initialization and are extremely sensitive to model initialization. The primary contribution of this paper is a scheme to automatically initialize an ASM for prostate segmentation on endorectal in vivo multi-protocol MRI via automated identification of MR spectra that lie within the prostate. A replicated clustering scheme is employed to distinguish prostatic from extra-prostatic MR spectra in the midgland. The spatial locations of the prostate spectra so identified are used as the initial ROI for a 2D ASM. The midgland initializations are used to define a ROI that is then scaled in 3D to cover the base and apex of the prostate. A multi-feature ASM employing statistical texture features is then used to drive the edge detection instead of just image intensity information alone. Quantitative comparison with another recent ASM initialization method by Cosio showed that our scheme resulted in a superior average segmentation performance on a total of 388 2D MRI sections obtained from 32 3D endorectal in vivo patient studies. Initialization of a 2D ASM via our MRS-based clustering scheme resulted in an average overlap accuracy (true positive ratio) of 0.60, while the scheme of Cosio yielded a corresponding average accuracy of 0.56 over 388 2D MR image sections. During an ASM segmentation, using no initialization resulted in an overlap of 0.53, using the Cosio based methodology resulted in an overlap of 0.60, and using the MRS-based methodology resulted in an overlap of 0.67, with a paired Student’s t-test indicating statistical significance to a high degree for all results. We also show that the final ASM segmentation result is highly correlated (as high as0.90) to the initialization scheme.

Introduction

Prostatic adenocarcinoma (CaP) is the second leading cause of cancer related deaths among men in the United States, with an estimated 186,000 new cases in 2008 (Source: American Cancer Society). The current standard for detection of CaP is transrectal ultrasound (TRUS) guided symmetrical needle biopsy, which has a high false negative rate associated with it (Catalona, 1991). Recently, multi-modal Magnetic Resonance (MR) Imaging (MRI) comprising both structural T2-weighted (T2-w) MRI (Madabhushi et al., 2005, Zhu et al., 2003) and MR Spectroscopy (MRS) (Kurhanewicz et al., 1996, Kurhanewicz et al., 2002, Kumar et al., 2008, Vilanova and Barcelo, 2007, Hom et al., 2006, Tiwari et al., 2009, Zaider et al., 2000, Kim et al., 2003, Coakley et al., 2003) have emerged as promising modalities for early detection of CaP (Kumar et al., 2008, Vilanova and Barcelo, 2007). MRS measures the relative concentrations of different biochemicals and metabolites in the prostate, and changes in relative concentrations of choline, creatine, and citrate are highly indicative of the presence of CaP. It is important to note that MRS acquisition has a lower resolution than MRI acquisition, and thus each MRS metavoxel (containing a spectral signal) is approximately 13 times the size of an MRI voxel (containing a single intensity value).

An example of a MR spectra signature associated with a T2-w MRI image is show in Fig. 1. The spectra corresponding to three metavoxels within the prostate are shown in red, and three spectra corresponding to metavoxels outside the prostate are shown in cyan. The average spectra of the extra-prostatic metavoxels is shown in Fig. 1h as a blue line, and the average spectra of the prostatic metavoxels is shown as a red line. It can be seen that the prostatic MRS spectra are greatly different from the extra-prostatic MRS spectra. Finally, in Fig. 1i a scatter plot of the MRS spectra for a given slice is shown, in which the prostatic spectra are indicated by red dots and the extra-prostatic spectra are indicated by blue dots. To visualize the 256-dimensional spectra in three dimensions, principal component analysis was used. This scatter plot shows an example of how the prostatic and extra-prostatic spectra are distinct.

As of 2009, there are approximately 16 ongoing clinical trials in the US aiming to demonstrate the role of MR in a diagnostic, clinical setting.¹ Recent literature suggests that the integration of MRI and MRS could potentially improve sensitivity and specificity for CaP detection (Hom et al., 2006). In fact, when combined with MRI, using MRS data could yield prostate cancer detection specificity and sensitivity values as high as 90% and 88% respectively (Testa et al., 2007). Recently, computer-aided diagnosis (CAD) schemes have emerged for automated CaP detection from prostate T2-w MRI (Madabhushi et al., 2005, Chan et al., 2003) and MRS (Tiwari et al., 2007, Tiwari et al., 2008, Tiwari et al., 2009). In Tiwari et al. (2009), we showed that spectral clustering of the MRS data could be used to distinguish between prostatic and extra-prostatic voxels with accuracies as high as 98%. This paper improves upon the methodology presented in Tiwari et al. (2009) to drive a segmentation scheme for the prostate capsule on T2-w MRI.

With the recent advancements of prostate MR imaging, several prostate segmentation schemes have been developed (Zhu et al., 2003, Chiu et al., 2004, Costa et al., 2007, Ladak et al., 2000, Hu et al., 2003, Pathak et al., 2000, Gong et al., 2004, Cosio, 2008, Gao et al., 2010). Segmentation of the prostate is useful for a number of tasks, including calculating the prostate volume pre- and post-treatment (Hoffelt et al., 2003, Kaminski et al., 2002), for creating patient specific anatomical models (Nathan et al., 1996), and for planning surgeries by helping to determine just how far outside the capsule they might need to go in order to capture any possible extra-capsular spread of the tumor. Additionally, identifying the prostate capsule is clinically significant for determining whether extra-capsular spread of CaP has occurred. Manual segmentation of the prostate, however, is not only laborious, but is also subject to a high degree of inter-, and intra-observer variability (Warfield et al., 2002, Warfield et al., 2004). The aim of this work is to automatically identify the spectra within the prostate in order to initialize a multi-feature active shape model (ASM) for precise segmentation of the prostate capsule.