Accurate prostate volume estimation using multifeature active shape models on T2-weighted MRI

The efficiency of prostate-specific antigen density measurement using three different methods on the prediction of biochemical recurrence

BACKGROUND

The aim of this study was to evaluate the efficiency of prostate-specific antigen (PSA) density (PSAD) calculated through prostate volume (PV) obtained via transrectal ultrasound (TRUS) and magnetic resonance imaging (MRI) and actual prostate weight (PW) methods obtained via pathological evaluation on the prediction of biochemical recurrence (BCR) in the follow-ups of patients who had undergone radical prostatectomy (RP).

METHODS

A total of 335 clinically localized prostate cancer (PCa) patients who had received open RP between January 2015 and December 2018 were enrolled in the study. Pre and postoperative demographic data, clinical and pathological findings and BCR conditions were recorded. The PSAD was calculated using information obtained through preoperative TRUS examinations, MRI, and collected pathological specimens after RP by dividing the maximum preoperative PSA value and PV/PW.

RESULTS

In a mean follow-up duration of 20.2 ± 8.5 months, recurrence was observed in 52 patients (24.4%) and progression was observed in 8 (3.8%) patients. The TRUS-PSAD, MRI-PSAD, and PW-PSAD values were statistically significantly higher in BCR patients compared to non-BCR patients. The International Society of Urologic Pathologists (ISUP) grade 5 and pT3b as a pathological stage were detected as independent variables in the prediction of BCR formation. Actual PW had a high prediction value compared to other PSAD measurements at <40 g prostate weights, but it had a low prediction value in prostates with an actual PW >60 g.

CONCLUSIONS

In this study, it was stated that PSAD acquired through different imaging methods does not affect the usability of PSAD in BCR prediction in clinical practice. The ISUP grade 5 and pT3b stage PCa were detected as independent markers in BCR prediction after RP.

Fully automated multiorgan segmentation of female pelvic magnetic resonance images with coarse-to-fine convolutional neural network

PURPOSE

Brachytherapy combined with external beam radiotherapy (EBRT) is the standard treatment for cervical cancer and has been shown to improve overall survival rates compared to EBRT only. Magnetic resonance (MR) imaging is used for radiotherapy (RT) planning and image guidance due to its excellent soft tissue image contrast. Rapid and accurate segmentation of organs at risk (OAR) is a crucial step in MR image-guided RT. In this paper, we propose a fully automated two-step convolutional neural network (CNN) approach to delineate multiple OARs from T2-weighted (T2W) MR images.

METHODS

We employ a coarse-to-fine segmentation strategy. The coarse segmentation step first identifies the approximate boundary of each organ of interest and crops the MR volume around the centroid of organ-specific region of interest (ROI). The cropped ROI volumes are then fed to organ-specific fine segmentation networks to produce detailed segmentation of each organ. A three-dimensional (3-D) U-Net is trained to perform the coarse segmentation. For the fine segmentation, a 3-D Dense U-Net is employed in which a modified 3-D dense block is incorporated into the 3-D U-Net-like network to acquire inter and intra-slice features and improve information flow while reducing computational complexity. Two sets of T2W MR images (221 cases for MR1 and 62 for MR2) were taken with slightly different imaging parameters and used for our network training and test. The network was first trained on MR1 which was a larger sample set. The trained model was then transferred to the MR2 domain via a fine-tuning approach. Active learning strategy was utilized for selecting the most valuable data from MR2 to be included in the adaptation via transfer learning.

RESULTS

The proposed method was tested on 20 MR1 and 32 MR2 test sets. Mean ± SD dice similarity coefficients are 0.93 ± 0.04, 0.87 ± 0.03, and 0.80 ± 0.10 on MR1 and 0.94 ± 0.05, 0.88 ± 0.04, and 0.80 ± 0.05 on MR2 for bladder, rectum, and sigmoid, respectively. Hausdorff distances (95th percentile) are 4.18 ± 0.52, 2.54 ± 0.41, and 5.03 ± 1.31 mm on MR1 and 2.89 ± 0.33, 2.24 ± 0.40, and 3.28 ± 1.08 mm on MR2, respectively. The performance of our method is superior to other state-of-the-art segmentation methods.

CONCLUSIONS

We proposed a two-step CNN approach for fully automated segmentation of female pelvic MR bladder, rectum, and sigmoid from T2W MR volume. Our experimental results demonstrate that the developed method is accurate, fast, and reproducible, and outperforms alternative state-of-the-art methods for OAR segmentation significantly (p < 0.05).

Comparison of PI-RADS Versions 2.0 and 2.1 for MRI-based Calculation of the Prostate Volume

RATIONALE AND OBJECTIVES

Prostate gland volume (PGV) should be routinely included in MRI reports of the prostate. The recently updated Prostate Imaging Reporting and Data System (PI-RADS) version 2.1 includes a change in the recommended measurement method for PGV compared to version 2.0. The purpose of this study was to evaluate the agreement of MRI-based PGV calculations with the volumetric manual slice-by-slice prostate segmentation as a reference standard using the linear measurements per PI-RADS versions 2.0 and 2.1. Furthermore, to assess inter-reader agreement for the different measurement approaches, determine the influence of an enlarged transition zone on measurement accuracy and to assess the value of the bullet formula for PGV calculation.

MATERIALS AND METHODS

Ninety-five consecutive treatment-naive patients undergoing prostate MRI were retrospectively analyzed. Prostates were manually contoured and segmented on axial T2-weighted images. Four different radiologists independently measured the prostate in three dimensions according to PI-RADS v2.0 and v2.1, respectively. MRI-based PGV was calculated using the ellipsoid and bullet formulas. Calculated volumes were compared to the reference manual segmentations using Wilcoxon signed-rank test. Inter-reader agreement was calculated using intraclass correlation coefficient (ICC).

RESULTS

Inter-reader agreement was excellent for the ellipsoid and bullet formulas using PI-RADS v2.0 (ICC 0.985 and 0.987) and v2.1 (ICC 0.990 and 0.994), respectively. The median difference from the reference standard using the ellipsoid formula derived PGV was 0.4 mL (interquartile range, -3.9 to 5.1 mL) for PI-RADS v2.0 (p = 0.393) and 2.6 mL (interquartile range, -1.6 to 7.3 mL) for v2.1 (p < 0.001) with a median difference of 2.2 mL. The bullet formula overestimated PGV by a median of 13.3 mL using PI-RADS v2.0 (p < 0.001) and 16.0 mL using v2.1 (p < 0.001). In the presence of an enlarged transition zone the PGV tended to be higher than the reference standard for PI-RADS v2.0 (median difference of 4.7 mL; p = 0.018) and for v2.1 (median difference of 5.7 mL, p < 0.001) using the ellipsoid formula.

CONCLUSION

Inter-reader agreement was excellent for the calculated PGV for both methods. PI-RADS v2.0 measurements with the ellipsoid formula yielded the most accurate volume estimates. The differences between PI-RADS v2.0 and v2.1 were statistically significant although small in absolute numbers but may be of relevance in specific clinical scenarios like prostate-specific antigen density calculation. These findings validate the use of the ellipsoid formula and highlight that the bullet formula should not be used for prostate volume estimation due to systematic overestimation.

The Comparison of Imaging and Clinical Methods to Estimate Prostate Volume: A Single-Centre Retrospective Study

BACKGROUND

Prostate volume (PV) is a useful tool in risk stratification, diagnosis, and follow-up of numerous prostatic diseases including prostate cancer and benign prostatic hypertrophy. There is currently no accepted ideal PV measurement method.

OBJECTIVE

This study compares multiple means of PV estimation, including digital rectal examination (DRE), transrectal ultrasound (TRUS), and magnetic resonance imaging (MRI), and radical prostatectomy specimens to determine the best volume measurement style.

METHODS

A retrospective, observational, single-site study with patients identified using an institutional database was performed. A total of 197 patients who underwent robot-assisted radical prostatectomy were considered. Data collected included age, serum PSA at the time of the prostate biopsy, clinical T stage, Gleason score, and PVs for each of the following methods: DRE, TRUS, MRI, and surgical specimen weight (SPW) and volume.

RESULTS

A paired t test was performed, which reported a statistically significant difference between PV measures (DRE, TRUS, MRI ellipsoid, MRI bullet, SP ellipsoid, and SP bullet) and the actual prostate weight. Lowest differences were reported for SP ellipsoid volume (M = -2.37; standard deviation [SD] = 10.227; t[167] = -3.011; and p = 0.003), MRI ellipsoid volume (M = -4.318; SD = 9.53; t[167] = -5.87; and p = 0.000), and MRI bullet volume (M = 5.31; SD = 10.77; t[167] = 6.387; and p = 0.000).

CONCLUSION

The PV obtained by MRI has proven to correlate with the PV obtained via auto-segmentation software as well as actual SPW, while also being more cost-effective and time-efficient. Therefore, demonstrating that MRI estimated the PV is an adequate method for use in clinical practice for therapeutic planning and patient follow-up.

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

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.

Which is the best radiological imaging method for predicting actual prostate weight?

In this study, we compared the weight of the prostate specimen removed after robotic radical prostatectomy with the prostate weight measured pre-operatively by four different imaging modalities. Pre-operative prostate weight before robotic radical prostatectomy was measured by Transabdominal Ultrasonography (TAUS), Transrectal Ultrasonography (TRUS), Abdominal Tomography (CT) and MultiparametricProstate Magnetic Resonance imaging (mpMRI). Of the 170 patients enrolled in the study, the mean age was 65.2 ± 7.08 (46-84) years and mean prostate-specific antigen (PSA) 9.6 ± 7.7 (1.8-50). The mean post-operative actual prostate weight was 63.1 ± 30 gr. The mean pre-operative prostate volumes measured by TAUS, TRUS, CT and MPMRI were 64.5 ± 28.5, 49.1 ± 30.6, 54.5 ± 30.5 and 68.7 ± 31.7 ml, respectively (p < .001). Post-operative actual prostate weight correlated with prostate weight measured by TAUS, TRUS, CT and mpMRI (r coefficient 0.776, 0.802, 0.768 and 0.825 respectively). The best of these was mpMRI. Although prostate weight measured by different imaging methods has a high correlation to predict actual prostate weight, actual prostate weight is best predicted by measurements with mpMRI. However, errors and deviations that may occur with these imaging methods should be taken into consideration.

Three-Dimensional Convolutional Neural Network for Prostate MRI Segmentation and Comparison of Prostate Volume Measurements by Use of Artificial Neural Network and Ellipsoid Formula

OBJECTIVE. The purposes of this study were to assess the performance of a 3D convolutional neural network (CNN) for automatic segmentation of prostates on MR images and to compare the volume estimates from the 3D CNN with those of the ellipsoid formula. MATERIALS AND METHODS. The study included 330 MR image sets that were divided into 260 training sets and 70 test sets for automated segmentation of the entire prostate. Among these, 162 training sets and 50 test sets were used for transition zone segmentation. Assisted by manual segmentation by two radiologists, the following values were obtained: estimates of ground-truth volume (V_GT), software-derived volume (V_SW), mean of V_GT and V_SW (V_AV), and automatically generated volume from the 3D CNN (V_NET). These values were compared with the volume calculated with the ellipsoid formula (V_EL). RESULTS. The Dice similarity coefficient for the entire prostate was 87.12% and for the transition zone was 76.48%. There was no significant difference between V_NET and V_AV (p = 0.689) in the test sets of the entire prostate, whereas a significant difference was found between V_EL and V_AV (p < 0.001). No significant difference was found among the volume estimates in the test sets of the transition zone. Overall intraclass correlation coefficients between the volume estimates were excellent (0.887-0.995). In the test sets of entire prostate, the mean error between V_GT and V_NET (2.5) was smaller than that between V_GT and V_EL (3.3). CONCLUSION. The fully automated network studied provides reliable volume estimates of the entire prostate compared with those obtained with the ellipsoid formula. Fast and accurate volume measurement by use of the 3D CNN may help clinicians evaluate prostate disease.

Radiomics and radiogenomics of prostate cancer

Radiomics and radiogenomics are attractive research topics in prostate cancer. Radiomics mainly focuses on extraction of quantitative information from medical imaging, whereas radiogenomics aims to correlate these imaging features to genomic data. The purpose of this review is to provide a brief overview summarizing recent progress in the application of radiomics-based approaches in prostate cancer and to discuss the potential role of radiogenomics in prostate cancer.

STRAINet: Spatially Varying sTochastic Residual AdversarIal Networks for MRI Pelvic Organ Segmentation

Accurate segmentation of pelvic organs is important for prostate radiation therapy. Modern radiation therapy starts to use a magnetic resonance image (MRI) as an alternative to computed tomography image because of its superior soft tissue contrast and also free of risk from radiation exposure. However, segmentation of pelvic organs from MRI is a challenging problem due to inconsistent organ appearance across patients and also large intrapatient anatomical variations across treatment days. To address such challenges, we propose a novel deep network architecture, called "Spatially varying sTochastic Residual AdversarIal Network" (STRAINet), to delineate pelvic organs from MRI in an end-to-end fashion. Compared to the traditional fully convolutional networks (FCN), the proposed architecture has two main contributions: 1) inspired by the recent success of residual learning, we propose an evolutionary version of the residual unit, i.e., stochastic residual unit, and use it to the plain convolutional layers in the FCN. We further propose long-range stochastic residual connections to pass features from shallow layers to deep layers; and 2) we propose to integrate three previously proposed network strategies to form a new network for better medical image segmentation: a) we apply dilated convolution in the smallest resolution feature maps, so that we can gain a larger receptive field without overly losing spatial information; b) we propose a spatially varying convolutional layer that adapts convolutional filters to different regions of interest; and c) an adversarial network is proposed to further correct the segmented organ structures. Finally, STRAINet is used to iteratively refine the segmentation probability maps in an autocontext manner. Experimental results show that our STRAINet achieved the state-of-the-art segmentation accuracy. Further analysis also indicates that our proposed network components contribute most to the performance.

A propagation-DNN: Deep combination learning of multi-level features for MR prostate segmentation

BACKGROUND AND OBJECTIVE

Prostate segmentation on Magnetic Resonance (MR) imaging is problematic because disease changes the shape and boundaries of the gland and it can be difficult to separate the prostate from surrounding tissues. We propose an automated model that extracts and combines multi-level features in a deep neural network to segment prostate on MR images.

METHODS

Our proposed model, the Propagation Deep Neural Network (P-DNN), incorporates the optimal combination of multi-level feature extraction as a single model. High level features from the convolved data using DNN are extracted for prostate localization and shape recognition, while labeling propagation, by low level cues, is embedded into a deep layer to delineate the prostate boundary.

RESULTS

A well-recognized benchmarking dataset (50 training data and 30 testing data from patients) was used to evaluate the P-DNN. When compared it to existing DNN methods, the P-DNN statistically outperformed the baseline DNN models with an average improvement in the DSC of 3.19%. When compared to the state-of-the-art non-DNN prostate segmentation methods, P-DNN was competitive by achieving 89.9 ± 2.8% DSC and 6.84 ± 2.5 mm HD on training sets and 84.13 ± 5.18% DSC and 9.74 ± 4.21 mm HD on testing sets.

CONCLUSION

Our results show that P-DNN maximizes multi-level feature extraction for prostate segmentation of MR images.

Determination of Prostate Volume: A Comparison of Contemporary Methods

RATIONALE AND OBJECTIVES

Prostate volume (PV) determination provides important clinical information. We compared PVs determined by digital rectal examination (DRE), transrectal ultrasound (TRUS), magnetic resonance imaging (MRI) with or without three-dimensional (3D) segmentation software, and surgical prostatectomy weight (SPW) and volume (SPV).

MATERIALS AND METHODS

This retrospective review from 2010 to 2016 included patients who underwent radical prostatectomy ≤1 year after multiparametric prostate MRI. PVs from DRE and TRUS were obtained from urology clinic notes. MRI-based PVs were calculated using bullet and ellipsoid formulas, automated 3D segmentation software (MRI-A3D), manual segmentation by a radiologist (MRI-R3D), and a third-year medical student (MRI-S3D). SPW and SPV were derived from pathology reports. Intraclass correlation coefficients compared the relative accuracy of each volume measurement.

RESULTS

Ninety-nine patients were analyzed. Median PVs were DRE 35 mL, TRUS 35 mL, MRI-bullet 49 mL, MRI-ellipsoid 39 mL, MRI-A3D 37 mL, MRI-R3D 36 mL, MRI-S3D 36 mL, SPW 54 mL, SPV-bullet 47 mL, and SPV-ellipsoid 37 mL. SPW and bullet formulas had consistently large PV, and formula-based PV had a wider spread than PV based on segmentation. Compared to MRI-R3D, the intraclass correlation coefficient was 0.91 for MRI-S3D, 0.90 for MRI-ellipsoid, 0.73 for SPV-ellipsoid, 0.72 for MRI-bullet, 0.71 for TRUS, 0.70 for SPW, 0.66 for SPV-bullet, 0.38 for MRI-A3D, and 0.33 for DRE.

CONCLUSIONS

With MRI-R3D measurement as the reference, the most reliable methods for PV estimation were MRI-S3D and MRI-ellipsoid formula. Automated segmentations must be individually assessed for accuracy, as they are not always truly representative of the prostate anatomy. Manual segmentation of the prostate does not require expert training.

SEMI-SUPERVISED LEARNING FOR PELVIC MR IMAGE SEGMENTATION BASED ON MULTI-TASK RESIDUAL FULLY CONVOLUTIONAL NETWORKS

Accurate segmentation of pelvic organs from magnetic resonance (MR) images plays an important role in image-guided radiotherapy. However, it is a challenging task due to inconsistent organ appearances and large shape variations. Fully convolutional network (FCN) has recently achieved state-of-the-art performance in medical image segmentation, but it requires a large amount of labeled data for training, which is usually difficult to obtain in real situation. To address these challenges, we propose a deep learning based semi-supervised learning framework. Specifically, we first train an initial multi-task residual fully convolutional network (FCN) based on a limited number of labeled MRI data. Based on the initially trained FCN, those unlabeled new data can be automatically segmented and some reasonable segmentations (after manual/automatic checking) can be included into the training data to fine-tune the network. This step can be repeated to progressively improve the training of our network, until no reasonable segmentations of new data can be included. Experimental results demonstrate the effectiveness of our proposed progressive semi-supervised learning fashion as well as its advantage in terms of accuracy.

Magnetic Resonance Imaging-Based Prostate-Specific Antigen Density for Prediction of Gleason Score Upgrade in Patients With Low-Risk Prostate Cancer on Initial Biopsy

OBJECTIVE

The aim of this study was to assess the utility of prostate-specific antigen density (PSAD) calculated using magnetic resonance imaging for predicting Gleason score (GS) upgrade in patients with low-risk prostate cancer on biopsy.

METHODS

Seventy-three patients were divided into 2 groups according to the concordance between biopsy and prostatectomy GS: group 1 (6/6) and group 2 (6/≥7). Magnetic resonance imaging-based PSAD, prostate volume, prostate-specific antigen (PSA), and age were compared between the 2 groups. Logistic regression and receiver operating characteristic curve analysis were performed.

RESULTS

Gleason score was upgraded in 40 patients. Patients in group 2 had significantly higher PSAD and PSA values and smaller prostate volume than did those in group 1. Prostate-specific antigen density of 0.26 ng/mL per cm or higher, PSA of 7.63 ng/mL or higher, and prostate volume of 25.1 cm or less were related to GS upgrade, with area-under-the-curve values of 0.765, 0.721, and 0.639, respectively.

CONCLUSIONS

Magnetic resonance imaging-based PSAD could help in predicting postoperative GS upgrade in patients with low-risk prostate cancer.

Automatic prostate segmentation on MR images with deep network and graph model

Automated prostate diagnoses and treatments have gained much attention due to the high mortality rate of prostate cancer. In particular, unsupervised (automatic) prostate segmentation is an active and challenging research. Most conventional works usually utilize handcrafted (low-level) features for prostate segmentation; however they often fail to extract the intrinsic structure of the prostate, especially on images with blurred boundaries. In this paper, we propose a novel automated prostate segmentation model with learned features from deep network. Specifically, we first generate a set of prostate proposals in transverse plane via recognizing the position and coarse estimate of the shape of the prostate on the global prostate image and using the deep network to extract highly effective features for the boundary refinement in a finer scale. With consideration of the correlations among different sequential images, we then construct a graph to select the best prostate proposals from proposal set for its use in 3D prostate segmentation. Experimental evaluation demonstrates that our proposed deep network and graph based method is superior to state-of-the-art couterparts, in terms of both dice similarity coefficient and Hausdorff distance, on public dataset.

Multi-atlas and unsupervised learning approach to perirectal space segmentation in CT images

Perirectal space segmentation in computed tomography images aids in quantifying radiation dose received by healthy tissues and toxicity during the course of radiation therapy treatment of the prostate. Radiation dose normalised by tissue volume facilitates predicting outcomes or possible harmful side effects of radiation therapy treatment. Manual segmentation of the perirectal space is time consuming and challenging in the presence of inter-patient anatomical variability and may suffer from inter- and intra-observer variabilities. However automatic or semi-automatic segmentation of the perirectal space in CT images is a challenging task due to inter patient anatomical variability, contrast variability and imaging artifacts. In the model presented here, a volume of interest is obtained in a multi-atlas based segmentation approach. Un-supervised learning in the volume of interest with a Gaussian-mixture-modeling based clustering approach is adopted to achieve a soft segmentation of the perirectal space. Probabilities from soft clustering are further refined by rigid registration of the multi-atlas mask in a probabilistic domain. A maximum a posteriori approach is adopted to achieve a binary segmentation from the refined probabilities. A mean volume similarity value of 97% and a mean surface difference of 3.06 ± 0.51 mm is achieved in a leave-one-patient-out validation framework with a subset of a clinical trial dataset. Qualitative results show a good approximation of the perirectal space volume compared to the ground truth.

Deformable MR Prostate Segmentation via Deep Feature Learning and Sparse Patch Matching

Automatic and reliable segmentation of the prostate is an important but difficult task for various clinical applications such as prostate cancer radiotherapy. The main challenges for accurate MR prostate localization lie in two aspects: (1) inhomogeneous and inconsistent appearance around prostate boundary, and (2) the large shape variation across different patients. To tackle these two problems, we propose a new deformable MR prostate segmentation method by unifying deep feature learning with the sparse patch matching. First, instead of directly using handcrafted features, we propose to learn the latent feature representation from prostate MR images by the stacked sparse auto-encoder (SSAE). Since the deep learning algorithm learns the feature hierarchy from the data, the learned features are often more concise and effective than the handcrafted features in describing the underlying data. To improve the discriminability of learned features, we further refine the feature representation in a supervised fashion. Second, based on the learned features, a sparse patch matching method is proposed to infer a prostate likelihood map by transferring the prostate labels from multiple atlases to the new prostate MR image. Finally, a deformable segmentation is used to integrate a sparse shape model with the prostate likelihood map for achieving the final segmentation. The proposed method has been extensively evaluated on the dataset that contains 66 T2-wighted prostate MR images. Experimental results show that the deep-learned features are more effective than the handcrafted features in guiding MR prostate segmentation. Moreover, our method shows superior performance than other state-of-the-art segmentation methods.

Comparison of prostate volume measured by endorectal coil MRI to prostate specimen volume and mass after radical prostatectomy

RATIONALE AND OBJECTIVES

To compare prostate volume measurements from 3-Tesla endorectal coil magnetic resonance imaging (ERC MRI) obtained with the prolate ellipsoid volume formula (EVF) and volumetry to pathology-based volume measurements.

METHODS

The institutional review board waived informed consent for this retrospective, health insurance portability and accountability act (HIPAA) compliant study, which included 195 patients who underwent 3-T ERC MRI between January 2008 and October 2011 and had pathologic prostate measurements available. Two readers in consensus measured the prostate length, height, and width on each MRI. They estimated prostate volumes using the prolate EVF (length × height × width × [π/6]) and also by performing three-dimensional volumetry. Pathologic specimen mass and dimensions were used to calculate prostate volume. Agreement was measured with Lin's concordance correlation coefficient (CCC). Volume differences were assessed using the Wilcoxon signed-rank test. Correct prostate-specific antigen (PSA) density classification rates were compared between EVF-based and volumetry-based PSA density levels using the exact McNemar test, with pathology-based PSA density as the reference standard.

RESULTS

Concordance was high between EVF and volumetry measurements (CCC, 0.950 [95% confidence interval, 0.935-0.962]) and between both kinds of MRI measurements and pathology (both CCC > 0.80). Based on a cut-off of ≤0.15 ng/mL/cm(3), use of EVF-based volume produced correct classification of 46 of 48 PSA density levels >15 ng/mL/cm(3) and 113 of 147 PSA density levels ≤15 ng/mL/cm(3); use of volumetry-based volume produced correct classification of 47 of 48 PSA density levels >15 ng/mL/cm(3) and 121 of 147 PSA density levels ≤15 ng/mL/cm(3). Rates of underclassification (P > .95) and overclassification (P = .10) did not differ significantly between EVF and volumetry.

CONCLUSIONS

EVF appears to be suitable for measuring prostate volume from ERC-MRI.

Deformable segmentation of 3D MR prostate images via distributed discriminative dictionary and ensemble learning

PURPOSE

Automatic prostate segmentation from MR images is an important task in various clinical applications such as prostate cancer staging and MR-guided radiotherapy planning. However, the large appearance and shape variations of the prostate in MR images make the segmentation problem difficult to solve. Traditional Active Shape/Appearance Model (ASM/AAM) has limited accuracy on this problem, since its basic assumption, i.e., both shape and appearance of the targeted organ follow Gaussian distributions, is invalid in prostate MR images. To this end, the authors propose a sparse dictionary learning method to model the image appearance in a nonparametric fashion and further integrate the appearance model into a deformable segmentation framework for prostate MR segmentation.

METHODS

To drive the deformable model for prostate segmentation, the authors propose nonparametric appearance and shape models. The nonparametric appearance model is based on a novel dictionary learning method, namely distributed discriminative dictionary (DDD) learning, which is able to capture fine distinctions in image appearance. To increase the differential power of traditional dictionary-based classification methods, the authors' DDD learning approach takes three strategies. First, two dictionaries for prostate and nonprostate tissues are built, respectively, using the discriminative features obtained from minimum redundancy maximum relevance feature selection. Second, linear discriminant analysis is employed as a linear classifier to boost the optimal separation between prostate and nonprostate tissues, based on the representation residuals from sparse representation. Third, to enhance the robustness of the authors' classification method, multiple local dictionaries are learned for local regions along the prostate boundary (each with small appearance variations), instead of learning one global classifier for the entire prostate. These discriminative dictionaries are located on different patches of the prostate surface and trained to adaptively capture the appearance in different prostate zones, thus achieving better local tissue differentiation. For each local region, multiple classifiers are trained based on the randomly selected samples and finally assembled by a specific fusion method. In addition to this nonparametric appearance model, a prostate shape model is learned from the shape statistics using a novel approach, sparse shape composition, which can model nonGaussian distributions of shape variation and regularize the 3D mesh deformation by constraining it within the observed shape subspace.

RESULTS

The proposed method has been evaluated on two datasets consisting of T2-weighted MR prostate images. For the first (internal) dataset, the classification effectiveness of the authors' improved dictionary learning has been validated by comparing it with three other variants of traditional dictionary learning methods. The experimental results show that the authors' method yields a Dice Ratio of 89.1% compared to the manual segmentation, which is more accurate than the three state-of-the-art MR prostate segmentation methods under comparison. For the second dataset, the MICCAI 2012 challenge dataset, the authors' proposed method yields a Dice Ratio of 87.4%, which also achieves better segmentation accuracy than other methods under comparison.

CONCLUSIONS

A new magnetic resonance image prostate segmentation method is proposed based on the combination of deformable model and dictionary learning methods, which achieves more accurate segmentation performance on prostate T2 MR images.

Evaluation of prostate segmentation algorithms for MRI: the PROMISE12 challenge

Prostate MRI image segmentation has been an area of intense research due to the increased use of MRI as a modality for the clinical workup of prostate cancer. Segmentation is useful for various tasks, e.g. to accurately localize prostate boundaries for radiotherapy or to initialize multi-modal registration algorithms. In the past, it has been difficult for research groups to evaluate prostate segmentation algorithms on multi-center, multi-vendor and multi-protocol data. Especially because we are dealing with MR images, image appearance, resolution and the presence of artifacts are affected by differences in scanners and/or protocols, which in turn can have a large influence on algorithm accuracy. The Prostate MR Image Segmentation (PROMISE12) challenge was setup to allow a fair and meaningful comparison of segmentation methods on the basis of performance and robustness. In this work we will discuss the initial results of the online PROMISE12 challenge, and the results obtained in the live challenge workshop hosted by the MICCAI2012 conference. In the challenge, 100 prostate MR cases from 4 different centers were included, with differences in scanner manufacturer, field strength and protocol. A total of 11 teams from academic research groups and industry participated. Algorithms showed a wide variety in methods and implementation, including active appearance models, atlas registration and level sets. Evaluation was performed using boundary and volume based metrics which were combined into a single score relating the metrics to human expert performance. The winners of the challenge where the algorithms by teams Imorphics and ScrAutoProstate, with scores of 85.72 and 84.29 overall. Both algorithms where significantly better than all other algorithms in the challenge (p<0.05) and had an efficient implementation with a run time of 8min and 3s per case respectively. Overall, active appearance model based approaches seemed to outperform other approaches like multi-atlas registration, both on accuracy and computation time. Although average algorithm performance was good to excellent and the Imorphics algorithm outperformed the second observer on average, we showed that algorithm combination might lead to further improvement, indicating that optimal performance for prostate segmentation is not yet obtained. All results are available online at http://promise12.grand-challenge.org/.

Simultaneous Segmentation of Prostatic Zones Using Active Appearance Models With Multiple Coupled Levelsets

In this work we present an improvement to the popular Active Appearance Model (AAM) algorithm, that we call the Multiple-Levelset AAM (MLA). The MLA can simultaneously segment multiple objects, and makes use of multiple levelsets, rather than anatomical landmarks, to define the shapes. AAMs traditionally define the shape of each object using a set of anatomical landmarks. However, landmarks can be difficult to identify, and AAMs traditionally only allow for segmentation of a single object of interest. The MLA, which is a landmark independent AAM, allows for levelsets of multiple objects to be determined and allows for them to be coupled with image intensities. This gives the MLA the flexibility to simulataneously segmentation multiple objects of interest in a new image. In this work we apply the MLA to segment the prostate capsule, the prostate peripheral zone (PZ), and the prostate central gland (CG), from a set of 40 endorectal, T2-weighted MRI images. The MLA system we employ in this work leverages a hierarchical segmentation framework, so constructed as to exploit domain specific attributes, by utilizing a given prostate segmentation to help drive the segmentations of the CG and PZ, which are embedded within the prostate. Our coupled MLA scheme yielded mean Dice accuracy values of .81, .79 and .68 for the prostate, CG, and PZ, respectively using a leave-one-out cross validation scheme over 40 patient studies. When only considering the midgland of the prostate, the mean DSC values were .89, .84, and .76 for the prostate, CG, and PZ respectively.

Automated prostate segmentation in whole-body MRI scans for epidemiological studies

The whole prostatic volume (PV) is an important indicator for benign prostate hyperplasia. Correlating the PV with other clinical parameters in a population-based prospective cohort study (SHIP-2) requires valid prostate segmentation in a large number of whole-body MRI scans. The axial proton density fast spin echo fat saturated sequence is used for prostate screening in SHIP-2. Our automated segmentation method is based on support vector machines (SVM). We used three-dimensional neighborhood information to build classification vectors from automatically generated features and randomly selected 16 MR examinations for validation. The Hausdorff distance reached a mean value of 5.048 ± 2.413, and a mean value of 5.613 ± 2.897 compared to manual segmentation by observers A and B. The comparison between volume measurement of SVM-based segmentation and manual segmentation of observers A and B depicts a strong correlation resulting in Spearman's rank correlation coefficients (ρ) of 0.936 and 0.859, respectively. Our automated methodology based on SVM for prostate segmentation can segment the prostate in WBI scans with good segmentation quality and has considerable potential for integration in epidemiological studies.

A survey of prostate segmentation methodologies in ultrasound, magnetic resonance and computed tomography images

Prostate segmentation is a challenging task, and the challenges significantly differ from one imaging modality to another. Low contrast, speckle, micro-calcifications and imaging artifacts like shadow poses serious challenges to accurate prostate segmentation in transrectal ultrasound (TRUS) images. However in magnetic resonance (MR) images, superior soft tissue contrast highlights large variability in shape, size and texture information inside the prostate. In contrast poor soft tissue contrast between prostate and surrounding tissues in computed tomography (CT) images pose a challenge in accurate prostate segmentation. This article reviews the methods developed for prostate gland segmentation TRUS, MR and CT images, the three primary imaging modalities that aids prostate cancer diagnosis and treatment. The objective of this work is to study the key similarities and differences among the different methods, highlighting their strengths and weaknesses in order to assist in the choice of an appropriate segmentation methodology. We define a new taxonomy for prostate segmentation strategies that allows first to group the algorithms and then to point out the main advantages and drawbacks of each strategy. We provide a comprehensive description of the existing methods in all TRUS, MR and CT modalities, highlighting their key-points and features. Finally, a discussion on choosing the most appropriate segmentation strategy for a given imaging modality is provided. A quantitative comparison of the results as reported in literature is also presented.

Patient specific prostate segmentation in 3-d magnetic resonance images

Accurate localization of the prostate and its surrounding tissue is essential in the treatment of prostate cancer. This paper presents a novel approach to fully automatically segment the prostate, including its seminal vesicles, within a few minutes of a magnetic resonance (MR) scan acquired without an endorectal coil. Such MR images are important in external beam radiation therapy, where using an endorectal coil is highly undesirable. The segmentation is obtained using a deformable model that is trained on-the-fly so that it is specific to the patient's scan. This case specific deformable model consists of a patient specific initialized triangulated surface and image feature model that are trained during its initialization. The image feature model is used to deform the initialized surface by template matching image features (via normalized cross-correlation) to the features of the scan. The resulting deformations are regularized over the surface via well established simple surface smoothing algorithms, which is then made anatomically valid via an optimized shape model. Mean and median Dice's similarity coefficients (DSCs) of 0.85 and 0.87 were achieved when segmenting 3T MR clinical scans of 50 patients. The median DSC result was equal to the inter-rater DSC and had a mean absolute surface error of 1.85 mm. The approach is showed to perform well near the apex and seminal vesicles of the prostate.

Multifeature landmark-free active appearance models: application to prostate MRI segmentation

Active shape models (ASMs) and active appearance models (AAMs) are popular approaches for medical image segmentation that use shape information to drive the segmentation process. Both approaches rely on image derived landmarks (specified either manually or automatically) to define the object's shape, which require accurate triangulation and alignment. An alternative approach to modeling shape is the levelset representation, defined as a set of signed distances to the object's surface. In addition, using multiple image derived attributes (IDAs) such as gradient information has previously shown to offer improved segmentation results when applied to ASMs, yet little work has been done exploring IDAs in the context of AAMs. In this work, we present a novel AAM methodology that utilizes the levelset implementation to overcome the issues relating to specifying landmarks, and locates the object of interest in a new image using a registration based scheme. Additionally, the framework allows for incorporation of multiple IDAs. Our multifeature landmark-free AAM (MFLAAM) utilizes an efficient, intuitive, and accurate algorithm for identifying those IDAs that will offer the most accurate segmentations. In this paper, we evaluate our MFLAAM scheme for the problem of prostate segmentation from T2-w MRI volumes. On a cohort of 108 studies, the levelset MFLAAM yielded a mean Dice accuracy of 88% ± 5%, and a mean surface error of 1.5 mm ±.8 mm with a segmentation time of 150/s per volume. In comparison, a state of the art AAM yielded mean Dice and surface error values of 86% ± 9% and 1.6 mm ± 1.0 mm, respectively. The differences with respect to our levelset-based MFLAAM model are statistically significant . In addition, our results were in most cases superior to several recent state of the art prostate MRI segmentation methods.

Automated computer-derived prostate volumes from MR imaging data: comparison with radiologist-derived MR imaging and pathologic specimen volumes

PURPOSE

To compare prostate gland volume (PV) estimation of automated computer-generated multifeature active shape models (MFAs) performed with 3-T magnetic resonance (MR) imaging with that of other methods of PV assessment, with pathologic specimens as the reference standard.

MATERIALS AND METHODS

All subjects provided written informed consent for this HIPAA-compliant and institutional review board-approved study. Freshly weighed prostatectomy specimens from 91 patients (mean age, 59 years; range, 42-84 years) served as the reference standard. PVs were manually calculated by two independent readers from MR images by using the standard ellipsoid formula. Planimetry PV was calculated from gland areas generated by two independent investigators by using manually drawn regions of interest. Computer-automated assessment of PV with an MFA was determined by the aggregate computer-calculated prostate area over the range of axial T2-weighted prostate MR images. Linear regression, linear mixed-effects models, concordance correlation coefficients, and Bland-Altman limits of agreement were used to compare volume estimation methods.

RESULTS

MFA-derived PVs had the best correlation with pathologic specimen PVs (slope, 0.888). Planimetry derived volumes produced slopes of 0.864 and 0.804 for two independent readers when compared with specimen PVs. Ellipsoid formula-derived PVs had slopes closest to one when compared with planimetry PVs. Manual MR imaging and MFA PV estimates had high concordance correlation coefficients with pathologic specimens.

CONCLUSION

MFAs with axial T2-weighted MR imaging provided an automated and efficient tool with which to assess PV. Both MFAs and MR imaging planimetry require adjustments for optimized PV accuracy when compared with prostatectomy specimens.