Dice Similarity Coefficient Values Research Articles

Generative Adversarial Network (GAN) for Automatic Reconstruction of the 3D Spine Structure by Using Simulated Bi-Planar X-ray Images.

In this study, we modified the previously proposed X2CT-GAN to build a 2Dto3D-GAN of the spine. This study also incorporated the radiologist’s perspective in the adjustment of input signals to prove the feasibility of the automatic production of three-dimensional (3D) structures of the spine from simulated bi-planar two-dimensional (2D) X-ray images. Data from 1012 computed tomography (CT) studies of 984 patients were retrospectively collected. We tested this model under different dataset sizes (333, 666, and 1012) with different bone signal conditions to observe the training performance. A 10-fold cross-validation and five metrics—Dice similarity coefficient (DSC) value, Jaccard similarity coefficient (JSC), overlap volume (OV), and structural similarity index (SSIM)—were applied for model evaluation. The optimal mean values for DSC, JSC, OV, SSIM_anteroposterior (AP), and SSIM_Lateral (Lat) were 0.8192, 0.6984, 0.8624, 0.9261, and 0.9242, respectively. There was a significant improvement in the training performance under empirically enhanced bone signal conditions and with increasing training dataset sizes. These results demonstrate the potential of the clinical implantation of GAN for automatic production of 3D spine images from 2D images. This prototype model can serve as a foundation in future studies applying transfer learning for the development of advanced medical diagnostic techniques.

A lightweight segmentation method based on residual U-Net for MR images

Automatic segmentation of Magnetic Resonance Imaging (MRI), which bases on Residual U-Net (ResU-Net), helps radiologists to quickly assess the condition. However, the ResU-Net structure requires a large number of parameters and storage model space. It is not convenient to apply to mobile MRI device. To solve this problem, Depthwise Separable Convolution and Squeeze-and-Excitation Residual U-Networks (DSRU-Net) is proposed to segment MRI. Squeeze-and-Excitation method is a channel attention mechanism. The proposed method is conducive to simplify ResU-Net model, making ResU-Net more convenient to be applied to mobile MRI device. The fuzzy comprehensive evaluation method, which includes three evaluation factors are that the required parameters of the model, the value of Dice Similarity Coefficient (DSC), and the value of Hausdorff Distance (HD), is used to evaluate the test results of the proposed method on the MICCAI 2012 Prostate MR Image Segmentation (PROMISE12) challenge dataset and Automatic Cardiac Diagnosis Challenge (ACDC) dataset. The fuzzy comprehensive evaluation values obtained by the proposed method in 5 PROMISE12 samples and 15 ACDC samples are 0.9889 and 0.9652, respectively. Combining the average results of the two datasets, the proposed method has the best effect in balancing the accuracy of segmentation and the amount of model parameters.

Joint optic disk and cup segmentation for glaucoma screening using a region-based deep learning network.

To develop and validate an end-to-end region-based deep convolutional neural network (R-DCNN) to jointly segment the optic disc (OD) and optic cup (OC) in retinal fundus images for precise cup-to-disc ratio (CDR) measurement and glaucoma screening. In total, 2440 retinal fundus images were retrospectively obtained from 2033 participants. An R-DCNN was presented for joint OD and OC segmentation, where the OD and OC segmentation problems were formulated into object detection problems. We compared R-DCNN's segmentation performance on our in-house dataset with that of four ophthalmologists while performing quantitative, qualitative and generalization analyses on the publicly available both DRISHIT-GS and RIM-ONE v3 datasets. The Dice similarity coefficient (DC), Jaccard coefficient (JC), overlapping error (E), sensitivity (SE), specificity (SP) and area under the curve (AUC) were measured. On our in-house dataset, the proposed model achieved a 98.51% DC and a 97.07% JC for OD segmentation, and a 97.63% DC and a 95.39% JC for OC segmentation, achieving a performance level comparable to that of the ophthalmologists. On the DRISHTI-GS dataset, our approach achieved 97.23% and 94.17% results in DC and JC results for OD segmentation, respectively, while it achieved a 94.56% DC and an 89.92% JC for OC segmentation. Additionally, on the RIM-ONE v3 dataset, our model generated DC and JC values of 96.89% and 91.32% on the OD segmentation task, respectively, whereas the DC and JC values acquired for OC segmentation were 88.94% and 78.21%, respectively. The proposed approach achieved very encouraging performance on the OD and OC segmentation tasks, as well as in glaucoma screening. It has the potential to serve as a useful tool for computer-assisted glaucoma screening.

Segmentation and classification of brain tumors from MRI images based on adaptive mechanisms and ELDP feature descriptor

Generally, soft tissue information of the brain is analyzed using Magnetic Resonance Imaging (MRI). An unusual growth of cells (tumor) in or around the brain may affect its overall functionality. Among brain tumors, gliomas that appear frequently in adults pose various challenges to radiologists. If they are not recognized properly in the initial phases, patients may face severe health issues including death. So, early and precise diagnosis of gliomas is significant for increasing the survival rate of patients. Recently, several computer-aided diagnosis (CAD) approaches have been developed for the detection and classification of gliomas. However, it is a challenging issue due to the noise as well as low sensitive boundary pixels appearing in gliomas. Considering this idea, we present a new approach based on adaptive methodologies and local texture descriptors. Here, primarily, we apply a median filter to remove noise from brain MRI images. Then, we employ adaptive decomposition strategies, namely bi-dimensional empirical mode decomposition (BEMD), and modified quasi-bivariate variational mode decomposition (MQBVMD) to attain meaningful sub-images. Later, we extract hidden texture details from the sub-images using an ELDP feature descriptor and then use a support vector machine (SVM) to distinguish brain MRI images as low-grade (LG) and high-grade (HG). Finally, we identify the infected region of gliomas with the help of adaptive K-means clustering and morphological operations. Using these sequences of operations, the proposed framework obtained 94.44% classification accuracy and 83.14% of dice similarity coefficient (DSC) value which is relatively high when compared to existing approaches.

Transfer Learning-Based Autosegmentation of Primary Tumor Volumes of Glioblastomas Using Preoperative MRI for Radiotherapy Treatment.

ObjectivesGlioblastoma is the most common primary malignant brain tumor in adults and can be treated with radiation therapy. However, tumor target contouring for head radiation therapy is labor-intensive and highly dependent on the experience of the radiation oncologist. Recently, autosegmentation of the tumor target has been playing an increasingly important role in the development of radiotherapy plans. Therefore, we established a deep learning model and improved its performance in autosegmenting and contouring the primary gross tumor volume (GTV) of glioblastomas through transfer learning.MethodsThe preoperative MRI data of 20 patients with glioblastomas were collected from our department (ST) and split into a training set and testing set. We fine-tuned a deep learning model for autosegmentation of the hippocampus on separate MRI scans (RZ) through transfer learning and trained this deep learning model directly using the training set. Finally, we evaluated the performance of both trained models in autosegmenting glioblastomas using the testing set.ResultsThe fine-tuned model converged within 20 epochs, compared to over 50 epochs for the model trained directly by the same training set, and demonstrated better autosegmentation performance [Dice similarity coefficient (DSC) 0.9404 ± 0.0117, 95% Hausdorff distance (95HD) 1.8107 mm ±0.3964mm, average surface distance (ASD) 0.6003 mm ±0.1287mm] than the model trained directly (DSC 0.9158±0.0178, 95HD 2.5761 mm ± 0.5365mm, ASD 0.7579 mm ± 0.1468mm) with the same test set. The DSC, 95HD, and ASD values of the two models were significantly different (P<0.05).ConclusionA model developed with semisupervised transfer learning and trained on independent data achieved good performance in autosegmenting glioblastoma. The autosegmented volume of glioblastomas is sufficiently accurate for radiotherapy treatment, which could have a positive impact on tumor control and patient survival.

BUSIS: A Benchmark for Breast Ultrasound Image Segmentation.

Breast ultrasound (BUS) image segmentation is challenging and critical for BUS computer-aided diagnosis (CAD) systems. Many BUS segmentation approaches have been studied in the last two decades, but the performances of most approaches have been assessed using relatively small private datasets with different quantitative metrics, which results in a discrepancy in performance comparison. Therefore, there is a pressing need for building a benchmark to compare existing methods using a public dataset objectively, to determine the performance of the best breast tumor segmentation algorithm available today, and to investigate what segmentation strategies are valuable in clinical practice and theoretical study. In this work, a benchmark for B-mode breast ultrasound image segmentation is presented. In the benchmark, (1) we collected 562 breast ultrasound images and proposed standardized procedures to obtain accurate annotations using four radiologists; (2) we extensively compared the performance of 16 state-of-the-art segmentation methods and demonstrated that most deep learning-based approaches achieved high dice similarity coefficient values (DSC ≥ 0.90) and outperformed conventional approaches; (3) we proposed the losses-based approach to evaluate the sensitivity of semi-automatic segmentation to user interactions; and (4) the successful segmentation strategies and possible future improvements were discussed in details.

Evaluation Exploration of Atlas-Based and Deep Learning-Based Automatic Contouring for Nasopharyngeal Carcinoma.

PurposeThe purpose of this study was to evaluate and explore the difference between an atlas-based and deep learning (DL)-based auto-segmentation scheme for organs at risk (OARs) of nasopharyngeal carcinoma cases to provide valuable help for clinical practice.Methods120 nasopharyngeal carcinoma cases were established in the MIM Maestro (atlas) database and trained by a DL-based model (AccuContour®), and another 20 nasopharyngeal carcinoma cases were randomly selected outside the atlas database. The experienced physicians contoured 14 OARs from 20 patients based on the published consensus guidelines, and these were defined as the reference volumes (Vref). Meanwhile, these OARs were auto-contoured using an atlas-based model, a pre-built DL-based model, and an on-site trained DL-based model. These volumes were named Vatlas, VDL-pre-built, and VDL-trained, respectively. The similarities between Vatlas, VDL-pre-built, VDL-trained, and Vref were assessed using the Dice similarity coefficient (DSC), Jaccard coefficient (JAC), maximum Hausdorff distance (HDmax), and deviation of centroid (DC) methods. A one-way ANOVA test was carried out to show the differences (between each two of them).ResultsThe results of the three methods were almost similar for the brainstem and eyes. For inner ears and temporomandibular joints, the results of the pre-built DL-based model are the worst, as well as the results of atlas-based auto-segmentation for the lens. For the segmentation of optic nerves, the trained DL-based model shows the best performance (p < 0.05). For the contouring of the oral cavity, the DSC value of VDL-pre-built is the smallest, and VDL-trained is the most significant (p < 0.05). For the parotid glands, the DSC of Vatlas is the minimum (about 0.80 or so), and VDL-pre-built and VDL-trained are slightly larger (about 0.82 or so). In addition to the oral cavity, parotid glands, and the brainstem, the maximum Hausdorff distances of the other organs are below 0.5 cm using the trained DL-based segmentation model. The trained DL-based segmentation method behaves well in the contouring of all the organs that the maximum average deviation of the centroid is no more than 0.3 cm.ConclusionThe trained DL-based segmentation performs significantly better than atlas-based segmentation for nasopharyngeal carcinoma, especially for the OARs with small volumes. Although some delineation results still need further modification, auto-segmentation methods improve the work efficiency and provide a level of help for clinical work.

Multiscale Local Enhancement Deep Convolutional Networks for the Automated 3D Segmentation of Gross Tumor Volumes in Nasopharyngeal Carcinoma: A Multi-Institutional Dataset Study.

PurposeAccurate segmentation of gross target volume (GTV) from computed tomography (CT) images is a prerequisite in radiotherapy for nasopharyngeal carcinoma (NPC). However, this task is very challenging due to the low contrast at the boundary of the tumor and the great variety of sizes and morphologies of tumors between different stages. Meanwhile, the data source also seriously affect the results of segmentation. In this paper, we propose a novel three-dimensional (3D) automatic segmentation algorithm that adopts cascaded multiscale local enhancement of convolutional neural networks (CNNs) and conduct experiments on multi-institutional datasets to address the above problems.Materials and MethodsIn this study, we retrospectively collected CT images of 257 NPC patients to test the performance of the proposed automatic segmentation model, and conducted experiments on two additional multi-institutional datasets. Our novel segmentation framework consists of three parts. First, the segmentation framework is based on a 3D Res-UNet backbone model that has excellent segmentation performance. Then, we adopt a multiscale dilated convolution block to enhance the receptive field and focus on the target area and boundary for segmentation improvement. Finally, a central localization cascade model for local enhancement is designed to concentrate on the GTV region for fine segmentation to improve the robustness. The Dice similarity coefficient (DSC), positive predictive value (PPV), sensitivity (SEN), average symmetric surface distance (ASSD) and 95% Hausdorff distance (HD95) are utilized as qualitative evaluation criteria to estimate the performance of our automated segmentation algorithm.ResultsThe experimental results show that compared with other state-of-the-art methods, our modified version 3D Res-UNet backbone has excellent performance and achieves the best results in terms of the quantitative metrics DSC, PPR, ASSD and HD95, which reached 74.49 ± 7.81%, 79.97 ± 13.90%, 1.49 ± 0.65 mm and 5.06 ± 3.30 mm, respectively. It should be noted that the receptive field enhancement mechanism and cascade architecture can have a great impact on the stable output of automatic segmentation results with high accuracy, which is critical for an algorithm. The final DSC, SEN, ASSD and HD95 values can be increased to 76.23 ± 6.45%, 79.14 ± 12.48%, 1.39 ± 5.44mm, 4.72 ± 3.04mm. In addition, the outcomes of multi-institution experiments demonstrate that our model is robust and generalizable and can achieve good performance through transfer learning.ConclusionsThe proposed algorithm could accurately segment NPC in CT images from multi-institutional datasets and thereby may improve and facilitate clinical applications.

H-ProSeg: Hybrid ultrasound prostate segmentation based on explainability-guided mathematical model

Background and ObjectiveAccurate and robust prostate segmentation in transrectal ultrasound (TRUS) images is of great interest for image-guided prostate interventions and prostate cancer diagnosis. However, it remains a challenging task for various reasons, including a missing or ambiguous boundary between the prostate and surrounding tissues, the presence of shadow artifacts, intra-prostate intensity heterogeneity, and anatomical variations. MethodsHere, we present a hybrid method for prostate segmentation (H-ProSeg) in TRUS images, using a small number of radiologist-defined seed points as the prior points. This method consists of three subnetworks. The first subnetwork uses an improved principal curve-based model to obtain data sequences consisting of seed points and their corresponding projection index. The second subnetwork uses an improved differential evolution-based artificial neural network for training to decrease the model error. The third subnetwork uses the parameters of the artificial neural network to explain the smooth mathematical description of the prostate contour. The performance of the H-ProSeg method was assessed in 55 brachytherapy patients using Dice similarity coefficient (DSC), Jaccard similarity coefficient (Ω), and accuracy (ACC) values. ResultsThe H-ProSeg method achieved excellent segmentation accuracy, with DSC, Ω, and ACC values of 95.8%, 94.3%, and 95.4%, respectively. Meanwhile, the DSC, Ω, and ACC values of the proposed method were as high as 93.3%, 91.9%, and 93%, respectively, due to the influence of Gaussian noise (standard deviation of Gaussian function, σ = 50). Although the σ increased from 10 to 50, the DSC, Ω, and ACC values fluctuated by a maximum of approximately 2.5%, demonstrating the excellent robustness of our method. ConclusionsHere, we present a hybrid method for accurate and robust prostate ultrasound image segmentation. The H-ProSeg method achieved superior performance compared with current state-of-the-art techniques. The knowledge of precise boundaries of the prostate is crucial for the conservation of risk structures. The proposed models have the potential to improve prostate cancer diagnosis and therapeutic outcomes.

Gross Tumor Volume Definition and Comparative Assessment for Esophageal Squamous Cell Carcinoma From 3D 18F-FDG PET/CT by Deep Learning-Based Method.

BackgroundThe accurate definition of gross tumor volume (GTV) of esophageal squamous cell carcinoma (ESCC) can promote precise irradiation field determination, and further achieve the radiotherapy curative effect. This retrospective study is intended to assess the applicability of leveraging deep learning-based method to automatically define the GTV from 3D 18F-FDG PET/CT images of patients diagnosed with ESCC.MethodsWe perform experiments on a clinical cohort with 164 18F-FDG PET/CT scans. The state-of-the-art esophageal GTV segmentation deep neural net is first employed to delineate the lesion area on PET/CT images. Afterwards, we propose a novel equivalent truncated elliptical cone integral method (ETECIM) to estimate the GTV value. Indexes of Dice similarity coefficient (DSC), Hausdorff distance (HD), and mean surface distance (MSD) are used to evaluate the segmentation performance. Conformity index (CI), degree of inclusion (DI), and motion vector (MV) are used to assess the differences between predicted and ground truth tumors. Statistical differences in the GTV, DI, and position are also determined.ResultsWe perform 4-fold cross-validation for evaluation, reporting the values of DSC, HD, and MSD as 0.72 ± 0.02, 11.87 ± 4.20 mm, and 2.43 ± 0.60 mm (mean ± standard deviation), respectively. Pearson correlations (R2) achieve 0.8434, 0.8004, 0.9239, and 0.7119 for each fold cross-validation, and there is no significant difference (t = 1.193, p = 0.235) between the predicted and ground truth GTVs. For DI, a significant difference is found (t = −2.263, p = 0.009). For position assessment, there is no significant difference (left-right in x direction: t = 0.102, p = 0.919, anterior–posterior in y direction: t = 0.221, p = 0.826, and cranial–caudal in z direction: t = 0.569, p = 0.570) between the predicted and ground truth GTVs. The median of CI is 0.63, and the gotten MV is small.ConclusionsThe predicted tumors correspond well with the manual ground truth. The proposed GTV estimation approach ETECIM is more precise than the most commonly used voxel volume summation method. The ground truth GTVs can be solved out due to the good linear correlation with the predicted results. Deep learning-based method shows its promising in GTV definition and clinical radiotherapy application.

A coronary artery CTA segmentation approach based on deep learning.

Presence of plaque and coronary artery stenosis are the main causes of coronary heart disease. Detection of plaque and coronary artery segmentation have become the first choice in detecting coronary artery disease. The purpose of this study is to investigate a new method for plaque detection and automatic segmentation and diagnosis of coronary arteries and to test its feasibility of applying to clinical medical image diagnosis. A multi-model fusion coronary CT angiography (CTA) vessel segmentation method is proposed based on deep learning. The method includes three network layer models namely, an original 3-dimensional full convolutional network (3D FCN) and two networks that embed the attention gating (AG) model in the original 3D FCN. Then, the prediction results of the three networks are merged by using the majority voting algorithm and thus the final prediction result of the networks is obtained. In the post-processing stage, the level set function is used to further iteratively optimize the results of network fusion prediction. The JI (Jaccard index) and DSC (Dice similarity coefficient) scores are calculated to evaluate accuracy of blood vessel segmentations. Applying to a CTA dataset of 20 patients, accuracy of coronary blood vessel segmentation using FCN, FCN-AG1, FCN-AG2 network and the fusion method are tested. The average values of JI and DSC of using the first three networks are (0.7962, 0.8843), (0.8154, 0.8966) and (0.8119, 0.8936), respectively. When using new fusion method, average JI and DSC of segmentation results increase to (0.8214, 0.9005), which are better than the best result of using FCN, FCN-AG1 and FCN-AG2 model independently.

Enhancement of evaluating flatfoot on a weight-bearing lateral radiograph of the foot with U-Net based semantic segmentation on the long axis of tarsal and metatarsal bones in an active learning manner

Robust labeling for semantic segmentation in radiographs is labor-intensive. No study has evaluated flatfoot-related deformities using semantic segmentation with U-Net on weight-bearing lateral radiographs. Here, we evaluated the robustness, accuracy enhancement, and efficiency of automated measurements for flatfoot-related angles using semantic segmentation in an active learning manner.A total of 300 consecutive weight-bearing lateral radiographs of the foot were acquired. The first 100 radiographs were used as the test set, and the following 200 radiographs were used as the training and validation sets, respectively. An expert orthopedic surgeon manually labeled ground truths. U-Net was used for model training. The Dice similarity coefficient (DSC) and Hausdorff distance (HD) were used to evaluate the segmentation results. In addition, angle measurement errors with a minimum moment of inertia (MMI) and ellipsoidal fitting (EF) based on the segmentation results were compared between active learning and learning with a pooled dataset.The mean values of DSC, HD, MMI, and EF of the average of all bones were 0.967, 1.274 mm, 0.792°, and 1.147° in active learning, and 0.964, 1.292 mm, 0.828°, and 1.186° in learning with a pooled dataset, respectively. The mean DSC and HD were significantly better in active learning than in learning with a pooled dataset. Labeling of all bones required 0.82 min in active learning and 0.88 min in learning with a pooled dataset.The accuracy and angle errors generally converged in both learning. However, the accuracies based on DSC and HD were significantly better in active learning. Moreover, active learning took less time for labeling, suggesting that active learning could be an accurate and efficient learning strategy for developing flatfoot classifiers based on semantic segmentation.

Computer-Aided Diagnosis for Determining Sagittal Spinal Curvatures Using Deep Learning and Radiography.

Analyzing spinal curvatures manually is time-consuming and tedious for clinicians, and intra-observer and inter-observer variability can affect manual measurements. In this study, we developed and evaluated the performance of an automated deep learning-based computer-aided diagnosis (CAD) tool for measuring the sagittal alignment of the spine from X-ray images. The CAD system proposed here performs two functions: deep learning-based lateral spine segmentation and automatic analysis of thoracic kyphosis and lumbar lordosis angles. We utilized 322 datasets with data augmentation for learning and fivefold cross-validation. The segmentation model was based on U-Net, which has multiple applications in medical image processing. Here, we utilized parameter equations and trigonometric functions to design spinal angle measurement algorithms. The kyphosis (T4-T12) and lordosis angle (L1-S1, L1-L5) were automatically measured to help diagnose kyphosis and lordosis. The segmentation model had precision, sensitivity, and dice similarity coefficient values of 90.53 ± 4.61%, 89.53 ± 1.8%, and 90.22 ± 0.62%, respectively. The performance of the CAD algorithm was also verified with the Pearson correlation, Bland-Altman, and intra-class correlation coefficient (ICC) analysis. The proposed angle measurement algorithm exhibited high similarity and reliability during verification. Therefore, CAD can help clinicians in reaching a diagnosis by analyzing the sagittal spinal curvatures while reducing observer-based variability and the required time or effort.

Learning a Discriminative Feature Attention Network for pancreas CT segmentation

Accurate pancreas segmentation is critical for the diagnosis and management of diseases of the pancreas. It is challenging to precisely delineate pancreas due to the highly variations in volume, shape and location. In recent years, coarse-to-fine methods have been widely used to alleviate class imbalance issue and improve pancreas segmentation accuracy. However, cascaded methods could be computationally intensive and the refined results are significantly dependent on the performance of its coarse segmentation results. To balance the segmentation accuracy and computational efficiency, we propose a Discriminative Feature Attention Network for pancreas segmentation, to effectively highlight pancreas features and improve segmentation accuracy without explicit pancreas location. The final segmentation is obtained by applying a simple yet effective post-processing step. Two experiments on both public NIH pancreas CT dataset and abdominal BTCV multi-organ dataset are individually conducted to show the effectiveness of our method for 2D pancreas segmentation. We obtained average Dice Similarity Coefficient (DSC) of 82.82±6.09%, average Jaccard Index (JI) of 71.13± 8.30% and average Symmetric Average Surface Distance (ASD) of 1.69 ± 0.83 mm on the NIH dataset. Compared to the existing deep learning-based pancreas segmentation methods, our experimental results achieve the best average DSC and JI value.

Development of Simplified Auto-Segmentable Functional Cardiac Atlas

PurposeThere is increasing evidence that radiation doses to cardiac substructures are associated with cardiac adverse events. Manual delineation of cardiac substructures is time-consuming, and auto-segmentation of cardiac substructure atlases has consequently been evaluated. However, proper automatic delineation of small substructures, such as the left anterior descending coronary artery, is challenging, and auto-segmentation of cardiac conduction system substructures has never been evaluated, despite multiple reports of radiation-induced arrhythmia after thoracic irradiations. The aim of this study was to propose and evaluate a simplified auto-segmentable functional cardiac atlas. Methods and MaterialsWe created a cardiac substructure atlas based on 20 computed tomography scans from patients with breast cancer comprising the 4 cardiac cavities, a high-risk cardiac zone as a left anterior descending coronary artery surrogate, and the 2 cardiac conduction nodes. Automatic delineation of this atlas by an atlas-based auto-segmentation algorithm was evaluated on a validation data set, consisting of 20 additional computed tomography scans. Dice similarity coefficients were used to evaluate the concordance level between the manual and the automatic contours; a dosimetric comparison between mean and maximum doses to the manual and to the auto-segmented substructures was additionally performed, based on intensity modulated radiation therapy treatment plans of the patients of the validation set. ResultsAverage dice similarity coefficient values were 0.78 for the 4 cardiac cavities, 0.65 for the high-risk cardiac zones, 0.56 for the sinoatrial node, and 0.15 for the atrioventricular node. Compared with manual contours, auto-segmented substructures were slightly smaller but the dosimetric parameters were similar. ConclusionsWe proposed a simplified functional cardiac atlas that included the cardiac conduction system and circumvented coronary delineation difficulties by using a surrogate high-risk cardiac zone. Most cardiac substructures were associated with acceptable atlas-based auto-segmentation properties. Such an atlas could be used for epidemiologic studies and for clinical practice.

A Deep Learning Image Data Augmentation Method for Single Tumor Segmentation.

PurposeMedical imaging examination is the primary method of diagnosis, treatment, and prevention of cancer. However, the amount of medical image data is often not enough to meet deep learning needs. This article aims to expand the small data set in tumor segmentation based on the deep learning method.MethodsThis method includes three main parts: image cutting and mirroring augmentation, segmentation of augmented images, and boundary reconstruction. Firstly, the image is divided into four parts horizontally & vertically, and diagonally along the tumor’s approximate center. Then each part is mirrored to get a new image and hence a four times data set. Next, the deep learning network trains the augmented data and gets the corresponding segmentation model. Finally, the segmentation boundary of the original tumor is obtained by boundary compensation and reconstruction.ResultsCombined with Mask-RCNN and U-Net, this study carried out experiments on a public breast ultrasound data set. The results show that the dice similarity coefficient (DSC) value obtained by horizontal and vertical cutting and mirroring augmentation and boundary reconstruction improved by 9.66% and 12.43% compared with no data augmentation. Moreover, the DSC obtained by diagonal cutting and mirroring augmentation and boundary reconstruction method improved by 9.46% and 13.74% compared with no data augmentation. Compared with data augmentation methods (cropping, rotating, and mirroring), this method’s DSC improved by 4.92% and 12.23% on Mask-RCNN and U-Net.ConclusionCompared with the traditional methods, the proposed data augmentation method has better performance in single tumor segmentation.

Comprehensive Quantitative Evaluation of Variability in Magnetic Resonance-Guided Delineation of Oropharyngeal Gross Tumor Volumes and High-Risk Clinical Target Volumes: An R-IDEAL Stage 0 Prospective Study

Tumor and target volume manual delineation remains a challenging task in head and neck cancer radiation therapy. The purpose of this study was to conduct a multi-institutional evaluation of manual delineations of gross tumor volume (GTV), high-risk clinical target volume (CTV), parotids, and submandibular glands on treatment simulation magnetic resonance scans of patients with oropharyngeal cancer. We retrospectively collected pretreatment T1-weighted, T1-weighted with gadolinium contrast, and T2-weighted magnetic resonance imaging scans for 4 patients with oropharyngeal cancer under an institution review board-approved protocol. We provided the scans to 26 radiation oncologists from 7 international cancer centers that participated in this delineation study. We also provide the patients' clinical history and physical examination findings, along with a medical photographic image and radiologic results. We used both the Simultaneous Truth and Performance Level Estimation algorithm and pair-wise comparisons of the contours, using overlap/distance metrics. Lastly, to assess experience and CTV delineation institutional practices, we had participants complete a brief questionnaire. Large variability was measured between observers' delineations for GTVs and CTVs. The mean Dice similarity coefficient values across all physicians' delineations for GTVp, GTVn, CTVp, and CTVn were 0.77, 0.67, 0.77, and 0.69, respectively, for Simultaneous Truth and Performance Level Estimation algorithm comparison, and 0.67, 0.60, 0.67, and 0.58, respectively, for pair-wise analysis. Normal tissue contours were defined more consistently when considering overlap/distance metrics. The median radiation oncology clinical experience was 7 years. The median experience delineating on magnetic resonance imaging was 3.5 years. The GTV-to-CTV margin used was 10 mm for 6 of 7 participant institutions. One institution used 8 mm, and 3 participants (from 3 different institutions) used a margin of 5 mm. The data from this study suggests that appropriate guidelines, contouring quality assurance sessions, and training are still needed for the adoption of magnetic resonance-based treatment planning for head and neck cancers. Such efforts should play a critical role in reducing delineation variation and ensure standardization of target design across clinical practices.

Knowledge-based quality control of organ delineations in radiation therapy.

To reduce the likelihood of errors in organ delineations used for radiotherapy treatment planning, a knowledge-based quality control (KBQC) system, which discriminates between valid and anomalous delineations isdeveloped. The KBQC is comprised of a group-wise inference system and anomaly detection modules trained using historical priors from 296 locally advanced lung and prostate cancer patient computational tomographies (CTs). The inference system discriminates different organs based on shape, relational, and intensity features. For a given delineated image set, the inference system solves a combinatorial optimization problem that results in an organ group whose relational features follow those of the training set considering the posterior probabilities obtained from support vector machine (SVM), discriminant subspace ensemble (DSE), and artificial neural network (ANN) classifiers. These classifiers are trained on nonrelational features with a 10-fold cross-validation scheme. The anomaly detection module is a bank of ANN autoencoders, each corresponding with an organ, trained on nonrelational features. A heuristic rule detects anomalous organs that exceed predefined organ-specific tolerances for the feature reconstruction error and the classifier's posterior probabilities. Independent data sets with anomalous delineations were used to test the overall performance of the KBQC system. The anomalous delineations were manually manipulated, computer-generated, or propagated based on a transformation obtained by imperfect registrations. Both peer-review-based scoring system and shape similarity coefficient (DSC) were used to label regions of interest (ROIs) as normal or anomalous in two independent test cohorts. The accuracy of the classifiers was 99.8%, and the minimum per-class F1-scores were 0.99, 0.99, and 0.98 for SVM, DSE, and ANN, respectively. The group-wise inference system reduced the miss-classification likelihood for the test data set with anomalous delineations compared to each individual classifier and a fused classifier that used the average posterior probability of all classifiers. For 15 independent locally advanced lung patients, the system detected 79% of the anomalous ROIs. For 1320 auto-segmented abdominopelvic organs, the anomaly detection system identified anomalous delineations, which also had low Dice similarity coefficient values with respect to manually delineated organs in the training dataset. The KBQC system detected anomalous delineations with superior accuracy compared to classification methods that judge only based on posterior probabilities.

Deep active contours using locally controlled distance vector flow

Active contours model (ACM) has been extensively used in computer vision and image processing. In recent studies, convolutional neural networks (CNNs) have been combined with ACM replacing the user in the process of contour evolution and image segmentation to eliminate limitations associated with ACM dependence on energy functional parameters and initialization. However, prior studies did not aim for automatic initialization, which is addressed in this article. In addition to manual initialization, current methods are highly sensitive to the initial location and fail to delineate borders accurately. We propose a fully automatic image segmentation method to address problems of manual initialization, insufficient capture range, and poor convergence to boundaries, in addition to the problem of assignment of energy functional parameters. We train two CNNs, one of which generating ACM weighting parameters and the other generating a ground truth mask to extract distance transform (DT) and an initialization circle. DT is used to form a vector field pointing from each pixel of the image towards the closest ground truth boundary point. Vector magnitudes are equal to the Euclidean distance between each pixel and the closest ground truth boundary point. We evaluate our method on four publicly available datasets, including two building instance segmentation datasets, i.e., Vaihingen and Bing huts, and two mammography image datasets, INBreast and DDSM-BCRP. Our approach achieves state-of-the-art results in mean Intersection over Union (mIoU), Dice similarity coefficient and Boundary F-score (BoundF) with the values of 92.33%, 92.44%, and 86.57% for Vaihingen dataset, and 87.12%, 86.86%, and 66.91% for Bing huts dataset. We obtained the Dice similarity coefficient values of 94.23% and 90.89% for the INBreast and DDSM-BCRP, respectively.

ThoraxNet: a 3D U-Net based two-stage framework for OAR segmentation on thoracic CT images.

An important phase of radiation treatment planning is the accurate contouring of the organs at risk (OAR), which is necessary for the dose distribution calculation. The manual contouring approach currently used in clinical practice is tedious, time-consuming, and prone to inter and intra-observer variation. Therefore, a deep learning-based auto contouring tool can solve these issues by accurately delineating OARs on the computed tomography (CT) images. This paper proposes a two-stage deep learning-based segmentation model with an attention mechanism that automatically delineates OARs in thoracic CT images. After preprocessing the input CT volume, a 3D U-Net architecture will locate each organ to generate cropped images for the segmentation network. Next, two differently configured U-Net-based networks will perform the segmentation of large organs-left lung, right lung, heart, and small organs-esophagus and spinal cord, respectively. A post-processing step integrates all the individually-segmented organs to generate the final result. The suggested model outperformed the state-of-the-art approaches in terms of dice similarity coefficient (DSC) values for the lungs and the heart. It is worth mentioning that the proposed model achieved a dice score of 0.941, which is 1.1% higher than the best previous dice score, in the case of the heart, an important organ in the human body. Moreover, the clinical acceptance of the results is verified using dosimetric analysis. To delineate all five organs on a CT scan of size [Formula: see text], our model takes only 8.61s. The proposed open-source automatic contouring tool can generate accurate contours in minimal time, consequently speeding up the treatment time and reducing the treatment cost.