Search

Abstract The diagnosis of early prostate cancer depends on the accurate segmentation of prostate regions in magnetic resonance imaging (MRI). However, this segmentation task is challenging due to the particularities of prostate MR images themselves and the limitations of existing methods. To address these issues, we propose a U-shaped encoder-decoder network MM-UNet based on Mamba and CNN for prostate segmentation in MR images. Specifically, we first proposed an adaptive feature fusion module based on channel attention guidance to achieve effective fusion between adjacent hierarchical features and suppress the interference of background noise. Secondly, we propose a global context-aware module based on Mamba, which has strong long-range modeling capabilities and linear complexity, to capture global context information in images. Finally, we propose a multi-scale anisotropic convolution module based on the principle of parallel multi-scale anisotropic convolution blocks and 3D convolution decomposition. Experimental results on two public prostate MR image segmentation datasets demonstrate that the proposed method outperforms competing models in terms of prostate segmentation performance and achieves state-of-the-art performance. In future work, we intend to enhance the model's robustness and extend its applicability to additional medical image segmentation tasks.

Abstract Purpose objective(s) Manual contouring of the prostate region in planning computed tomography (CT) images is a challenging task due to factors such as low contrast in soft tissues, inter- and intra-observer variability, and variations in organ size and shape. Consequently, the use of automated contouring methods can offer significant advantages. In this study, we aimed to investigate automated male pelvic multi-organ contouring in multi-center planning CT images using a hybrid convolutional neural network-vision transformer (CNN-ViT) that combines convolutional and ViT techniques. Materials/methods We used retrospective data from 104 localized prostate cancer patients, with delineations of the clinical target volume (CTV) and critical organs at risk (OAR) for external beam radiotherapy. We introduced a novel attention-based fusion module that merges detailed features extracted through convolution with the global features obtained through the ViT. Results The average dice similarity coefficients (DSCs) achieved by VGG16-UNet-ViT for the prostate, bladder, rectum, right femoral head (RFH), and left femoral head (LFH) were 91.75%, 95.32%, 87.00%, 96.30%, and 96.34%, respectively. Experiments conducted on multi-center planning CT images indicate that combining the ViT structure with the CNN network resulted in superior performance for all organs compared to pure CNN and transformer architectures. Furthermore, the proposed method achieves more precise contours compared to state-of-the-art techniques. Conclusion Results demonstrate that integrating ViT into CNN architectures significantly improves segmentation performance. These results show promise as a reliable and efficient tool to facilitate prostate radiotherapy treatment planning.

Magnetic Resonance Imaging is increasing in importance in prostate cancer diagnosis due to the high accuracy and quality of the examination procedure. However, this process requires a time-consuming analysis of the results. Currently, machine vision is widely used in many areas. It enables automation and support in radiological studies. Successful detection of primary prostate tumors depends on the effective segmentation of the prostate itself. At times, a CT scan may be performed; alternatively, MRI may be the selected option. The data always reach a bottleneck stage. This paper presents the effective training of deep learning models to segment the prostate based on onefold and multimodal medical images. This approach supports the computer-aided diagnosis (CAD) system for radiologists as the first step in cancer exams. A comparison of two approaches designed for prostate segmentation is described. The first combines YOLOv4, the object detection neural network, and U-Net for a semantic segmentation based on onefold modality MRI images. The second presents the same method trained on multimodal images—a CT and MRI mixed dataset. The learning process was carried out in a cloud environment using GPU cards. The experiments are based on data from 120 patients who have undergone MRI and CT examinations. Several metrics evaluated the trained models. In the prostate semantic segmentation process, better results were achieved by mixed MRI with CT datasets. The best model achieved the value of 0.9685 for the Sørensen–Dice coefficient for the threshold value of 0.6.

Prostate gland segmentation is the primary step to estimate gland volume, which aids in the prostate disease management. In this study, we present a 2D-3D convolutional neural network (CNN) ensemble that automatically segments the whole prostate gland along with the peripheral zone (PZ) (PPZ-SegNet) using a T2-weighted sequence (T2W) of Magnetic Resonance Imaging (MRI). The study used 4 different public data sets organized as Train #1 and Test #1 (independently derived from the same cohort), Test #2, Test #3 and Test #4. The prostate gland and the peripheral zone (PZ) anatomy were manually delineated with consensus read by a radiologist, except for Test #4 cohorts that had pre-marked glandular anatomy. A Bayesian hyperparameter optimization method was applied to construct the network model (PPZ-SegNet) with a training cohort (Train #1, n = 150) using a five-fold cross validation. The model evaluation was performed on an independent cohort of 283 T2W MRI prostate cases (Test #1 to #4) without any additional tuning. The data cohorts were derived from The Cancer Imaging Archives (TCIA): PROSTATEx Challenge, Prostatectomy, Repeatability studies and PROMISE12-Challenge. The segmentation performance was evaluated by computing the Dice similarity coefficient and Hausdorff distance between the estimated-deep-network identified regions and the radiologist-drawn annotations. The deep network architecture was able to segment the prostate gland anatomy with an average Dice score of 0.86 in Test #1 (n = 192), 0.79 in Test #2 (n = 26), 0.81 in Test #3 (n = 15), and 0.62 in Test #4 (n = 50). We also found the Dice coefficient improved with larger prostate volumes in 3 of the 4 test cohorts. The variation of the Dice scores from different cohorts of test images suggests the necessity of more diverse models that are inclusive of dependencies such as the gland sizes and others, which will enable us to develop a universal network for prostate and PZ segmentation. Our training and evaluation code can be accessed through the link: https://github.com/mariabaldeon/PPZ-SegNet.git.

Segmentation of the prostate gland from magnetic resonance images is rapidly becoming a standard of care in prostate cancer radiotherapy treatment planning. Automating this process has the potential to improve accuracy and efficiency. However, the performance and accuracy of deep learning models varies depending on the design and optimal tuning of the hyper-parameters. In this study, we examine the effect of loss functions on the performance of deep-learning-based prostate segmentation models. A U-Net model for prostate segmentation using T2-weighted images from a local dataset was trained and performance compared when using nine different loss functions, including: Binary Cross-Entropy (BCE), Intersection over Union (IoU), Dice, BCE and Dice (BCE + Dice), weighted BCE and Dice (W (BCE + Dice)), Focal, Tversky, Focal Tversky, and Surface loss functions. Model outputs were compared using several metrics on a five-fold cross-validation set. Ranking of model performance was found to be dependent on the metric used to measure performance, but in general, W (BCE + Dice) and Focal Tversky performed well for all metrics (whole gland Dice similarity coefficient (DSC): 0.71 and 0.74; 95HD: 6.66 and 7.42; Ravid 0.05 and 0.18, respectively) and Surface loss generally ranked lowest (DSC: 0.40; 95HD: 13.64; Ravid −0.09). When comparing the performance of the models for the mid-gland, apex, and base parts of the prostate gland, the models’ performance was lower for the apex and base compared to the mid-gland. In conclusion, we have demonstrated that the performance of a deep learning model for prostate segmentation can be affected by choice of loss function. For prostate segmentation, it would appear that compound loss functions generally outperform singles loss functions such as Surface loss.

Prostate cancer is the most common cancer in men after lung cancer. Generally, the segmentation of the prostate is the preprocessing work for the diagnosis of prostate cancer. Aiming at the variety of prostate and the similarity of visual characteristics between prostates and their surroundings, this paper proposes a new prostate segmentation network based on MR images, denoted as ProSegNet. ProSegNet consists of two parts: encoder and decoder. To improve the feature extraction capability of the encoder, we use dense blocks as the feature extraction unit, and at the same time introduce a cross-stage partial (CSP) structure to reduce the amount of calculation. In the design of the decoder structure, we integrated the spatial attention mechanism and the channel attention mechanism to enable it to focus on the important features while ignoring the invalid features. In addition, to segment the prostate more accurately, we add a prostate contour segmentation branch to the output of the segmentation network to learn the contour features of the prostate. Finally, to alleviate the problem of small intensity difference between the prostate and surrounding tissues, we designed a truncated intensity stretching image enhancement method. The performance of ProSegNet has been experimentally verified in the Promise12 and ProstateX datasets. On the Promise12 dataset, the dice similarity coefficient (DSC) and hausdorff (Haus) distance are 0.908 and 9.87 respectively. On the ProstateX dataset, the DSC and Haus reach the results of 0.892 and 10.45, respectively. Experimental results show that the ProSegNet can obtain a competitive performance.

Prostate cancer is a malignant tumor that occurs in the male prostate. Prostate cancer lesions have the characteristics of small size and blurry outline, which is a challenge to design a robust prostate cancer detection method. At present, clinical diagnosis of prostate cancer is mainly based on magnetic resonance (MR) imaging. However, it is difficult to obtain prostate cancer data, and the data with true values is also very limited, which further increases the difficulty of prostate cancer detection methods based on MR images. To solve these problems, this paper designs a new method of prostate cancer detection based on MR images, which is recorded as ProCDet. The method consists of three modules: registration of prostate MR images, segmentation of prostate, and segmentation of prostate cancer lesions. First, the registration between different sequences of MR images is performed to find the spatial relationship between the different sequences. Then, the designed prostate segmentation network based on the attention mechanism is used to segment the prostate to remove the interference of background information. Finally, a 3D prostate cancer lesion segmentation network based on Focal Tversky Loss is applied to determine the specific location of prostate cancer. Moreover, in order to take full advantage of unlabeled prostate data, this paper designs a self-supervised learning method to improve the accuracy of prostate cancer detection. The proposed ProCDet has been experimentally verified on the ProstateX dataset. When the average number of false-positive lesions per patient is 0.6275, the true-positive rate is 91.82%. Experimental results show that the ProCDet can obtain competitive detection performance.

Whole-prostate gland (WPG) segmentation plays a significant role in prostate volume measurement, treatment, and biopsy planning. This study evaluated a previously developed automatic WPG segmentation, deep attentive neural network (DANN), on a large, continuous patient cohort to test its feasibility in a clinical setting. With IRB approval and HIPAA compliance, the study cohort included 3,698 3T MRI scans acquired between 2016 and 2020. In total, 335 MRI scans were used to train the model, and 3,210 and 100 were used to conduct the qualitative and quantitative evaluation of the model. In addition, the DANN-enabled prostate volume estimation was evaluated by using 50 MRI scans in comparison with manual prostate volume estimation. For qualitative evaluation, visual grading was used to evaluate the performance of WPG segmentation by two abdominal radiologists, and DANN demonstrated either acceptable or excellent performance in over 96% of the testing cohort on the WPG or each prostate sub-portion (apex, midgland, or base). Two radiologists reached a substantial agreement on WPG and midgland segmentation (κ = 0.75 and 0.63) and moderate agreement on apex and base segmentation (κ = 0.56 and 0.60). For quantitative evaluation, DANN demonstrated a dice similarity coefficient of 0.93 ± 0.02, significantly higher than other baseline methods, such as DeepLab v3+ and UNet (both p values < 0.05). For the volume measurement, 96% of the evaluation cohort achieved differences between the DANN-enabled and manual volume measurement within 95% limits of agreement. In conclusion, the study showed that the DANN achieved sufficient and consistent WPG segmentation on a large, continuous study cohort, demonstrating its great potential to serve as a tool to measure prostate volume.

Automatic and accurate segmentation of the prostate is still a challenging task due to intensity inhomogeneity and complicated deformation of MR images. To tackle these problems with multi-atlas segmentation, in this paper, we propose a new metric for image registration and new descriptor for label fusion. First, to reduce the amount of edges in entropic graph, a modified α-mutual information (α-MI) based on fast minimal spanning tree (MST) is implemented for deformable registration. Second, localized α-MI allowing for the spatial information is proposed with the stochastic gradient optimization, and the feature space is encoded by a sparse auto-encoder. Finally, a multi-scale descriptor utilizing local self-similarity is integrated into the patch-based label fusion to obtain final segmentation. Experiments were performed on two subsets of totally 46 T2-weighted prostate MR images from 46 patients. Compared to α-MI based on k-nearest neighbor graph, the registration time of α-MI based on fast MST can be reduced by almost half. The median Dice overlap of registration using localized α-MI on one subset is shown to improve significantly from 0.725 to 0.764 (p = 1.14 × 10-5), compared to using α-MI without the spatial information. The median Dice overlap of prostate segmentation using the proposed method on 20 testing images of another subset is 0.871, and the median Hausdorff distance is 8.013 mm, which demonstrate a comparable accuracy to state-of-the-art methods.

In prostate cancer, fusion biopsy, which couples magnetic resonance imaging (MRI) with transrectal ultrasound (TRUS), poses the basis for targeted biopsy by allowing the comparison of information coming from both imaging modalities at the same time. Compared with the standard clinical procedure, it provides a less invasive option for the patients and increases the likelihood of sampling cancerous tissue regions for the subsequent pathology analyses. As a prerequisite to image fusion, segmentation must be achieved from both MRI and TRUS domains. The automatic contour delineation of the prostate gland from TRUS images is a challenging task due to several factors including unclear boundaries, speckle noise, and the variety of prostate anatomical shapes. Automatic methodologies, such as those based on deep learning, require a huge quantity of training data to achieve satisfactory results. In this paper, the authors propose a novel optimization formulation to find the best superellipse, a deformable model that can accurately represent the prostate shape. The advantage of the proposed approach is that it does not require extensive annotations, and can be used independently of the specific transducer employed during prostate biopsies. Moreover, in order to show the clinical applicability of the method, this study also presents a module for the automatic segmentation of the prostate gland from MRI, exploiting the nnU-Net framework. Lastly, segmented contours from both imaging domains are fused with a customized registration algorithm in order to create a tool that can help the physician to perform a targeted prostate biopsy by interacting with the graphical user interface.

This paper proposes a method based on the active shape model (ASM) to segment the prostate in magnetic resonance (MR) images. Due to the great variability in appearance among different boundaries of the prostate and among subjects, the traditional ASM is weak in MR image prostate segmentation. To address these limitations, we investigated a novel ASM-based method by incorporating deep feature learning into our previous liver segmentation framework. First, an adaptive feature learning probability boosting tree (AFL-PBT) based on both simple handcrafted features and deep learned features was developed for prostate pre-segmentation and for further shape model initialization. The proposed AFL-PBT classifier also provided a boundary searching band, which made the ASM less sensitive to model initialization. Then, the convolutional neutral network (CNN) deep learning method was used to train a boundary model, which separated voxels into three types: near, inside, and outside the boundary. A three-level ASM based on the CNN boundary model was employed for the final segmentation refinement. On MICCAI PROMISE12 test data sets, the proposed method yielded a mean Dice score of 84% with a standard deviation of 4%. The experimental results demonstrated that the proposed method outperformed other ASM-based prostate MRI segmentation methods and achieved a level of accuracy comparable to that of state-of-the-art methods.

Prostate segmentations are required for an ever-increasing number of medical applications, such as image-based lesion detection, fusion-guided biopsy and focal therapies. However, obtaining accurate segmentations is laborious, requires expertise and, even then, the inter-observer variability remains high. In this paper, a robust, accurate and generalizable model for Magnetic Resonance (MR) and three-dimensional (3D) Ultrasound (US) prostate image segmentation is proposed. It uses a densenet-resnet-based Convolutional Neural Network (CNN) combined with techniques such as deep supervision, checkpoint ensembling and Neural Resolution Enhancement. The MR prostate segmentation model was trained with five challenging and heterogeneous MR prostate datasets (and two US datasets), with segmentations from many different experts with varying segmentation criteria. The model achieves a consistently strong performance in all datasets independently (mean Dice Similarity Coefficient -DSC- above 0.91 for all datasets except for one), outperforming the inter-expert variability significantly in MR (mean DSC of 0.9099 vs. 0.8794). When evaluated on the publicly available Promise12 challenge dataset, it attains a similar performance to the best entries. In summary, the model has the potential of having a significant impact on current prostate procedures, undercutting, and even eliminating, the need of manual segmentations through improvements in terms of robustness, generalizability and output resolution.

In this paper, we develop an optimised state-of-the-art 2D U-Net model by studying the effects of the individual deep learning model components in performing prostate segmentation. We found that for upsampling, the combination of interpolation and convolution is better than the use of transposed convolution. For combining feature maps in each convolution block, it is only beneficial if a skip connection with concatenation is used. With respect to pooling, average pooling is better than strided-convolution, max, RMS or L2 pooling. Introducing a batch normalisation layer before the activation layer gives further performance improvement. The optimisation is based on a private dataset as it has a fixed 2D resolution and voxel size for every image which mitigates the need of a resizing operation in the data preparation process. Non-enhancing data preprocessing was applied and five-fold cross-validation was used to evaluate the fully automatic segmentation approach. We show it outperforms the traditional methods that were previously applied on the private dataset, as well as outperforming other comparable state-of-the-art 2D models on the public dataset PROMISE12.