Segmentation of prostate zones using probabilistic atlas-based method with diffusion-weighted MR images

Highlights

• A robust probabilistic atlas-based approach is presented for semi-automatic segmentation of prostate using diffusion-weighted imaging.

• Partial volume correction algorithm is developed for precise estimation of the same probability level pixels between both zones.

• The segmentation framework aids the radiologists for accurate extraction of peripheral zone and transition zone, which can be combined in the computer-aided diagnostic system for prostate cancer.

Abstract

Background and objective

Accurate segmentation of prostate and its zones constitute an essential preprocessing step for computer-aided diagnosis and detection system for prostate cancer (PCa) using diffusion-weighted imaging (DWI). However, low signal-to-noise ratio and high variability of prostate anatomic structures are challenging for its segmentation using DWI. We propose a semi-automated framework that segments the prostate gland and its zones simultaneously using DWI.

Methods

In this paper, the Chan-Vese active contour model along with morphological opening operation was used for segmentation of prostate gland. Then segmentation of prostate zones into peripheral zone (PZ) and transition zone (TZ) was carried out using in-house developed probabilistic atlas with partial volume (PV) correction algorithm. The study cohort included MRI dataset of 18 patients (n = 18) as our dataset and methodology were also independently evaluated using 15 MRI scans (n = 15) of QIN-PROSTATE-Repeatability dataset. The atlas for zones of prostate gland was constructed using dataset of twelve patients of our patient cohort. Three-fold cross-validation was performed with 10 repetitions, thus total 30 instances of training and testing were performed on our dataset followed by independent testing on the QIN-PROSTATE-Repeatability dataset. Dice similarity coefficient (DSC), Jaccard coefficient (JC), and accuracy were used for quantitative assessment of the segmentation results with respect to boundaries delineated manually by an expert radiologist. A paired t-test was performed to evaluate the improvement in zonal segmentation performance with the proposed PV correction algorithm.

Results

For our dataset, the proposed segmentation methodology produced improved segmentation with DSC of 90.76 ± 3.68%, JC of 83.00 ± 5.78%, and accuracy of 99.42 ± 0.36% for the prostate gland, DSC of 77.73 ± 2.76%, JC of 64.46 ± 3.43%, and accuracy of 82.47 ± 2.22% for the PZ, and DSC of 86.05 ± 1.50%, JC of 75.80 ± 2.10%, and accuracy of 91.67 ± 1.56% for the TZ. The segmentation performance for QIN-PROSTATE-Repeatability dataset was, DSC of 85.50 ± 4.43%, JC of 75.00 ± 6.34%, and accuracy of 81.52 ± 5.55% for prostate gland, DSC of 74.40 ± 1.79%, JC of 59.53 ± 8.70%, and accuracy of 80.91 ± 5.16% for PZ, and DSC of 85.80 ± 5.55%, JC of 74.87 ± 7.90%, and accuracy of 90.59 ± 3.74% for TZ. With the implementation of the PV correction algorithm, statistically significant (p<0.05) improvements were observed in all the metrics (DSC, JC, and accuracy) for both prostate zones, PZ and TZ segmentation.

Conclusions

The proposed segmentation methodology is stable, accurate, and easy to implement for segmentation of prostate gland and its zones (PZ and TZ). The atlas-based segmentation framework with PV correction algorithm can be incorporated into a computer-aided diagnostic system for PCa localization and treatment planning.

Introduction

Prostate cancer (PCa) is the second most common cancer and the fifth leading cause of cancer-related death among men worldwide [1]. Accurate segmentation of prostate gland and its zones on MR images is a challenging task due to the high variability in prostate anatomic structures, lack of clear boundaries in-between its zones, and similar intensity profiles with nearby tissues [2].

Multi-parametric MRI analysis (mpMRI) uses various MRI sequences such as T2-weighted (T2W), diffusion-weighted imaging (DWI), and dynamic contrast-enhanced (DCE) imaging for evaluation of pathological lesions in prostate. mpMRI has significantly improved the diagnostic accuracy of PCa and has the potential to reduce number of unnecessary biopsies [3]. Prostate consists of three zones: central zone (CZ), transition zone (TZ) and peripheral zone (PZ). CZ and TZ are usually not clearly distinguishable on MR images, recently a study by Vargas et al. [4] showed that in patients with PCa undergoing MR imaging, the CZ was visible at least partially, in only 81-84% of patients. However, these two zones are commonly referred to combined as TZ for the purpose of image processing and analysis [5]. About 70-75% of PCa originate in PZ, and 20-30% in TZ [6]. An internationally accepted scoring and interpretation scheme, Prostate Imaging-Reporting and Data System version 2 (PI-RADS v2) has defined the evaluation criteria for the assessment of PZ and TZ separately because tumor exhibit different characteristics in these two zones [7]. It is important to thus sub-segment the prostate into its zones for any detailed clinical interpretation and quantitative analysis.

Although the accuracy for detection of PCa using mpMRI is good [8], it can be further improved using a reliable segmentation technique along with mpMRI. Manual delineation of tissues on each slice of 3D-MRI acquisition is a tedious and time-consuming task that is also highly subjective, which significantly increases the burden for radiologists. Therefore, an automatic, accurate and reproducible segmentation algorithm can reduce image interpretation time and represent a significant advancement in computer-aided diagnostic (CAD) system and possibly also in treatment planning.

Several segmentation algorithms for the prostate gland have been reported in literature which include region-based approach [9], contour and shape-based approach [10], supervised and unsupervised classification based approach [11,12] and hybrid of these methods [13], [14], [15]. However, only few studies have focused on the sub-segmentation of the prostate into its zones [16], [17], [18] and more recently deep learning has been applied using convolution neural network (CNN) based segmentation approaches [12,[19], [20], [21]]. Most segmentation work has been presented using T2W MR images [18], [19], [20], [21] and only two studies have used DWI [12,22]. PI-RADS v2 recommends DWI sequence as the most important for the interpretation of lesions in PZ and also a key feature in TZ lesions detection [6]. Therefore, using DWI for the segmentation process is beneficial, which might be useful in CAD algorithms of PCa to make the process suitable for clinical use. However, automated segmentation of prostate gland and its zones using DWI is a challenging task due to low signal-to-noise ratio (SNR).

In this paper, the Chan-Vese active contour model [23] along with morphological opening operation has been used for whole prostate gland segmentation, and a probabilistic atlas-based approach has been developed for prostate zonal segmentation. A zone is generated after matching the test subject with the atlas at similar probability level. In this zone, a pixel may contain both types of tissues PZ or TZ; this is called partial volume (PV) zone. For the accurate estimation of PV zone pixels, either belonging to PZ or TZ, a PV correction algorithm has been developed in this study. The aim of this study was to develop a semi-automated and accurate framework to segment prostate gland and its zones simultaneously using DWI.

The main structure of this paper is organized as follows. Methodology is presented in Section 2, the MRI data acquisition in Section 2.1, and data processing in Section 2.2. Prostate gland segmentation method is presented in Section 2.2.1 and atlas-based sub-segmentation of prostate zones in Section 2.2.2. Experimental results are elaborated in Section 3. Discussion and conclusion are presented in Sections 4 and 5.