Uncertainty Aware Temporal-Ensembling Model for Semi-Supervised ABUS Mass Segmentation

EEG-Based Cross-Dataset Driver Drowsiness Recognition With an Entropy Optimization Network

Cross-dataset driver drowsiness recognition with EEG is important for the advancement of a calibration-free driver drowsiness recognition system. Nevertheless, this task is challenging due to the impact of distribution drift on recognition accuracy. In this paper, we propose a novel model named entropy optimization network (EON) for the task. The model takes a novel two-step strategy to separate the unlabeled data from the target domain. It firstly uses a novel modified entropy loss to encourage unlabeled samples well aligned with the source domain to form clear clusters. Next, it gradually separates samples from the target domain with a self-training framework by taking adequate advantage of underlying patterns inherent in it. The proposed method is tested on the domain adaptation task with two public datasets and achieves 2-class recognition accuracies of and , which beats other baseline methods. Our work illuminates a promising direction in achieving the ultimate objective of developing a driver drowsiness recognition system without calibration.

A nomogram for diagnosis of BI-RADS 4 breast nodules based on three-dimensional volume ultrasound

OBJECTIVES

The classification of malignant breast nodules into four categories according to the Breast Imaging Reporting and Data System (BI-RADS) presents significant variability, posing challenges in clinical diagnosis. This study investigates whether a nomogram prediction model incorporating automated breast ultrasound system (ABUS) can improve the accuracy of differentiating benign and malignant BI-RADS 4 breast nodules.

METHODS

Data were collected for a total of 257 nodules with breast nodules corresponding to BI-RADS 4 who underwent ABUS examination and for whom pathology results were obtained from January 2019 to August 2022. The participants were divided into a benign group (188 cases) and a malignant group (69 cases) using a retrospective study method. Ultrasound imaging features were recorded. Logistic regression analysis was used to screen the clinical and ultrasound characteristics. Using the results of these analyses, a nomogram prediction model was established accordingly.

RESULTS

Age, distance between nodule and nipple, calcification and C-plane convergence sign were independent risk factors that enabled differentiation between benign and malignant breast nodules (all P < 0.05). A nomogram model was established based on these variables. The area under curve (AUC) values for the nomogram model, age, distance between nodule and nipple, calcification, and C-plane convergence sign were 0.86, 0.735, 0.645, 0.697, and 0.685, respectively. Thus, the AUC value for the model was significantly higher than a single variable.

CONCLUSIONS

A nomogram based on the clinical and ultrasound imaging features of ABUS can be used to improve the accuracy of the diagnosis of benign and malignant BI-RADS 4 nodules. It can function as a relatively accurate predictive tool for sonographers and clinicians and is therefore clinically useful. ADVANCES IN KNOWLEDGE STATEMENT: we retrospectively analyzed the clinical and ultrasound characteristics of ABUS BI-RADS 4 nodules and established a nomogram model to improve the efficiency of the majority of ABUS readers in the diagnosis of BI-RADS 4 nodules.

An accurate and trustworthy deep learning approach for bladder tumor segmentation with uncertainty estimation

BACKGROUND AND OBJECTIVE

Although deep learning-based intelligent diagnosis of bladder cancer has achieved excellent performance, the reliability of neural network predicted results may not be evaluated. This study aims to explore a trustworthy AI-based tumor segmentation model, which not only outputs predicted results but also provides confidence information about the predictions.

METHODS

This paper proposes a novel model for bladder tumor segmentation with uncertainty estimation (BSU), which is not merely able to effectively segment the lesion area but also yields an uncertainty map showing the confidence information of the segmentation results. In contrast to previous uncertainty estimation, we utilize test time augmentation (TTA) and test time dropout (TTD) to estimate aleatoric uncertainty and epistemic uncertainty in both internal and external datasets to explore the effects of both uncertainties on different datasets.

RESULTS

Our BSU model achieved the Dice coefficients of 0.766 and 0.848 on internal and external cystoscopy datasets, respectively, along with accuracy of 0.950 and 0.954. Compared to the state-of-the-art methods, our BSU model demonstrated superior performance, which was further validated by the statistically significance of the t-tests at the conventional level. Clinical experiments verified the practical value of uncertainty estimation in real-world bladder cancer diagnostics.

CONCLUSIONS

The proposed BSU model is able to visualize the confidence of the segmentation results, serving as a valuable addition for assisting urologists in enhancing both the precision and efficiency of bladder cancer diagnoses in clinical practice.

Semi-Supervised Burn Depth Segmentation Network with Contrast Learning and Uncertainty Correction

Burn injuries are a common traumatic condition, and the early diagnosis of burn depth is crucial for reducing treatment costs and improving survival rates. In recent years, image-based deep learning techniques have been utilized to realize the automation and standardization of burn depth segmentation. However, the scarcity and difficulty in labeling burn data limit the performance of traditional deep learning-based segmentation methods. Mainstream semi-supervised methods face challenges in burn depth segmentation due to single-level perturbations, lack of explicit edge modeling, and ineffective handling of inaccurate predictions in unlabeled data. To address these issues, we propose SBCU-Net, a semi-supervised burn depth segmentation network with contrastive learning and uncertainty correction. Building on the LTB-Net from our previous work, SBCU-Net introduces two additional decoder branches to enhance the consistency between the probability map and soft pseudo-labels under multi-level perturbations. To improve segmentation in complex regions like burn edges, contrastive learning refines the outputs of the three-branch decoder, enabling more discriminative feature representation learning. In addition, an uncertainty correction mechanism weights the consistency loss based on prediction uncertainty, reducing the impact of inaccurate pseudo-labels. Extensive experiments on burn datasets demonstrate that SBCU-Net effectively leverages unlabeled data and achieves superior performance compared to state-of-the-art semi-supervised methods.

Labeled-to-unlabeled distribution alignment for partially-supervised multi-organ medical image segmentation

Partially-supervised multi-organ medical image segmentation aims to develop a unified semantic segmentation model by utilizing multiple partially-labeled datasets, with each dataset providing labels for a single class of organs. However, the limited availability of labeled foreground organs and the absence of supervision to distinguish unlabeled foreground organs from the background pose a significant challenge, which leads to a distribution mismatch between labeled and unlabeled pixels. Although existing pseudo-labeling methods can be employed to learn from both labeled and unlabeled pixels, they are prone to performance degradation in this task, as they rely on the assumption that labeled and unlabeled pixels have the same distribution. In this paper, to address the problem of distribution mismatch, we propose a labeled-to-unlabeled distribution alignment (LTUDA) framework that aligns feature distributions and enhances discriminative capability. Specifically, we introduce a cross-set data augmentation strategy, which performs region-level mixing between labeled and unlabeled organs to reduce distribution discrepancy and enrich the training set. Besides, we propose a prototype-based distribution alignment method that implicitly reduces intra-class variation and increases the separation between the unlabeled foreground and background. This can be achieved by encouraging consistency between the outputs of two prototype classifiers and a linear classifier. Extensive experimental results on the AbdomenCT-1K dataset and a union of four benchmark datasets (including LiTS, MSD-Spleen, KiTS, and NIH82) demonstrate that our method outperforms the state-of-the-art partially-supervised methods by a considerable margin, and even surpasses the fully-supervised methods. The source code is publicly available at LTUDA.

U-Net and Its Variants Based Automatic Tracking of Radial Artery in Ultrasonic Short-Axis Views: A Pilot Study

Background/Objectives: Radial artery tracking (RAT) in the short-axis view is a pivotal step for ultrasound-guided radial artery catheterization (RAC), which is widely employed in various clinical settings. To eliminate disparities and lay the foundations for automated procedures, a pilot study was conducted to explore the feasibility of U-Net and its variants in automatic RAT. Methods: Approved by the institutional ethics committee, patients as potential RAC candidates were enrolled, and the radial arteries were continuously scanned by B-mode ultrasonography. All acquired videos were processed into standardized images, and randomly divided into training, validation, and test sets in an 8:1:1 ratio. Deep learning models, including U-Net and its variants, such as Attention U-Net, UNet++, Res-UNet, TransUNet, and UNeXt, were utilized for automatic RAT. The performance of the deep learning architectures was assessed using loss functions, dice similarity coefficient (DSC), and Jaccard similarity coefficient (JSC). Performance differences were analyzed using the Kruskal-Wallis test. Results: The independent datasets comprised 7233 images extracted from 178 videos of 135 patients (53.3% women; mean age: 41.6 years). Consistent convergence of loss functions between the training and validation sets was achieved for all models except Attention U-Net. Res-UNet emerged as the optimal architecture in terms of DSC and JSC (93.14% and 87.93%), indicating a significant improvement compared to U-Net (91.79% vs. 86.19%, p < 0.05) and Attention U-Net (91.20% vs. 85.02%, p < 0.05). Conclusions: This pilot study validates the feasibility of U-Net and its variants in automatic RAT, highlighting the predominant performance of Res-UNet among the evaluated architectures.

A review of uncertainty quantification in medical image analysis: Probabilistic and non-probabilistic methods

The comprehensive integration of machine learning healthcare models within clinical practice remains suboptimal, notwithstanding the proliferation of high-performing solutions reported in the literature. A predominant factor hindering widespread adoption pertains to an insufficiency of evidence affirming the reliability of the aforementioned models. Recently, uncertainty quantification methods have been proposed as a potential solution to quantify the reliability of machine learning models and thus increase the interpretability and acceptability of the results. In this review, we offer a comprehensive overview of the prevailing methods proposed to quantify the uncertainty inherent in machine learning models developed for various medical image tasks. Contrary to earlier reviews that exclusively focused on probabilistic methods, this review also explores non-probabilistic approaches, thereby furnishing a more holistic survey of research pertaining to uncertainty quantification for machine learning models. Analysis of medical images with the summary and discussion on medical applications and the corresponding uncertainty evaluation protocols are presented, which focus on the specific challenges of uncertainty in medical image analysis. We also highlight some potential future research work at the end. Generally, this review aims to allow researchers from both clinical and technical backgrounds to gain a quick and yet in-depth understanding of the research in uncertainty quantification for medical image analysis machine learning models.

Mine Your Own Anatomy: Revisiting Medical Image Segmentation With Extremely Limited Labels

Recent studies on contrastive learning have achieved remarkable performance solely by leveraging few labels in the context of medical image segmentation. Existing methods mainly focus on instance discrimination and invariant mapping (i.e., pulling positive samples closer and negative samples apart in the feature space). However, they face three common pitfalls: (1) tailness: medical image data usually follows an implicit long-tail class distribution. Blindly leveraging all pixels in training hence can lead to the data imbalance issues, and cause deteriorated performance; (2) consistency: it remains unclear whether a segmentation model has learned meaningful and yet consistent anatomical features due to the intra-class variations between different anatomical features; and (3) diversity: the intra-slice correlations within the entire dataset have received significantly less attention. This motivates us to seek a principled approach for strategically making use of the dataset itself to discover similar yet distinct samples from different anatomical views. In this paper, we introduce a novel semi-supervised 2D medical image segmentation framework termed Mine yOur owNAnatomy (MONA), and make three contributions. First, prior work argues that every pixel equally matters to the model training; we observe empirically that this alone is unlikely to define meaningful anatomical features, mainly due to lacking the supervision signal. We show two simple solutions towards learning invariances-through the use of stronger data augmentations and nearest neighbors. Second, we construct a set of objectives that encourage the model to be capable of decomposing medical images into a collection of anatomical features in an unsupervised manner. Lastly, we both empirically and theoretically, demonstrate the efficacy of our MONA on three benchmark datasets with different labeled settings, achieving new state-of-the-art under different labeled semi-supervised settings. MONA makes minimal assumptions on domain expertise, and hence constitutes a practical and versatile solution in medical image analysis. We provide the PyTorch-like pseudo-code in supplementary.

URCA: Uncertainty-based region clipping algorithm for semi-supervised medical image segmentation

BACKGROUND AND OBJECTIVE

Training convolutional neural networks based on large amount of labeled data has made great progress in the field of image segmentation. However, in medical image segmentation tasks, annotating the data is expensive and time-consuming because pixel-level annotation requires experts in the relevant field. Currently, the combination of consistent regularization and pseudo labeling-based semi-supervised methods has shown good performance in image segmentation. However, in the training process, a portion of low-confidence pseudo labels are generated by the model. And the semi-supervised segmentation method still has the problem of distribution bias between labeled and unlabeled data. The objective of this study is to address the challenges of semi-supervised learning and improve the segmentation accuracy of semi-supervised models on medical images.

METHODS

To address these issues, we propose an Uncertainty-based Region Clipping Algorithm for semi-supervised medical image segmentation, which consists of two main modules. A module is introduced to compute the uncertainty of two sub-networks predictions with diversity using Monte Carlo Dropout, allowing the model to gradually learn from more reliable targets. To retain model diversity, we use different loss functions for different branches and use Non-Maximum Suppression in one of the branches. The other module is proposed to generate new samples by masking the low-confidence pixels in the original image based on uncertainty information. New samples are fed into the model to facilitate the model to generate pseudo labels with high confidence and enlarge the training data distribution.

RESULTS

Comprehensive experiments on the combination of two benchmarks ACDC and BraTS2019 show that our proposed model outperforms state-of-the-art methods in terms of Dice, HD95 and ASD. The results reach an average Dice score of 87.86 % and a HD95 score of 4.214 mm on ACDC dataset. For the brain tumor segmentation, the results reach an average Dice score of 84.79 % and a HD score of 10.13 mm.

CONCLUSIONS

Our proposed method improves the accuracy of semi-supervised medical image segmentation. Extensive experiments on two public medical image datasets including 2D and 3D modalities demonstrate the superiority of our model. The code is available at: https://github.com/QuintinDong/URCA.

Hybrid dual mean-teacher network with double-uncertainty guidance for semi-supervised segmentation of magnetic resonance images

Semi-supervised learning has made significant progress in medical image segmentation. However, existing methods primarily utilize information from a single dimensionality, resulting in sub-optimal performance on challenging magnetic resonance imaging (MRI) data with multiple segmentation objects and anisotropic resolution. To address this issue, we present a Hybrid Dual Mean-Teacher (HD-Teacher) model with hybrid, semi-supervised, and multi-task learning to achieve effective semi-supervised segmentation. HD-Teacher employs a 2D and a 3D mean-teacher network to produce segmentation labels and signed distance fields from the hybrid information captured in both dimensionalities. This hybrid mechanism allows HD-Teacher to utilize features from 2D, 3D, or both dimensions as needed. Outputs from 2D and 3D teacher models are dynamically combined based on confidence scores, forming a single hybrid prediction with estimated uncertainty. We propose a hybrid regularization module to encourage both student models to produce results close to the uncertainty-weighted hybrid prediction to further improve their feature extraction capability. Extensive experiments of binary and multi-class segmentation conducted on three MRI datasets demonstrated that the proposed framework could (1) significantly outperform state-of-the-art semi-supervised methods (2) surpass a fully-supervised VNet trained on substantially more annotated data, and (3) perform on par with human raters on muscle and bone segmentation task. Code will be available at https://github.com/ThisGame42/Hybrid-Teacher.

Attention decoupled contrastive learning for semi-supervised segmentation method based on data augmentation

Objective.Deep learning algorithms have demonstrated impressive performance by leveraging large labeled data. However, acquiring pixel-level annotations for medical image analysis, especially in segmentation tasks, is both costly and time-consuming, posing challenges for supervised learning techniques. Existing semi-supervised methods tend to underutilize representations of unlabeled data and handle labeled and unlabeled data separately, neglecting their interdependencies.Approach.To address this issue, we introduce the Data-Augmented Attention-Decoupled Contrastive model (DADC). This model incorporates an attention decoupling module and utilizes contrastive learning to effectively distinguish foreground and background, significantly improving segmentation accuracy. Our approach integrates an augmentation technique that merges information from both labeled and unlabeled data, notably boosting network performance, especially in scenarios with limited labeled data.Main results.We conducted comprehensive experiments on the automated breast ultrasound (ABUS) dataset and the results demonstrate that DADC outperforms existing segmentation methods in terms of segmentation performance.

3D Vascular Segmentation Supervised by 2D Annotation of Maximum Intensity Projection

Vascular structure segmentation plays a crucial role in medical analysis and clinical applications. The practical adoption of fully supervised segmentation models is impeded by the intricacy and time-consuming nature of annotating vessels in the 3D space. This has spurred the exploration of weakly-supervised approaches that reduce reliance on expensive segmentation annotations. Despite this, existing weakly supervised methods employed in organ segmentation, which encompass points, bounding boxes, or graffiti, have exhibited suboptimal performance when handling sparse vascular structure. To alleviate this issue, we employ maximum intensity projection (MIP) to decrease the dimensionality of 3D volume to 2D image for efficient annotation, and the 2D labels are utilized to provide guidance and oversight for training 3D vessel segmentation model. Initially, we generate pseudo-labels for 3D blood vessels using the annotations of 2D projections. Subsequently, taking into account the acquisition method of the 2D labels, we introduce a weakly-supervised network that fuses 2D-3D deep features via MIP to further improve segmentation performance. Furthermore, we integrate confidence learning and uncertainty estimation to refine the generated pseudo-labels, followed by fine-tuning the segmentation network. Our method is validated on five datasets (including cerebral vessel, aorta and coronary artery), demonstrating highly competitive performance in segmenting vessels and the potential to significantly reduce the time and effort required for vessel annotation. Our code is available at: https://github.com/gzq17/Weakly-Supervised-by-MIP.

Trustworthy clinical AI solutions: A unified review of uncertainty quantification in Deep Learning models for medical image analysis

The full acceptance of Deep Learning (DL) models in the clinical field is rather low with respect to the quantity of high-performing solutions reported in the literature. End users are particularly reluctant to rely on the opaque predictions of DL models. Uncertainty quantification methods have been proposed in the literature as a potential solution, to reduce the black-box effect of DL models and increase the interpretability and the acceptability of the result by the final user. In this review, we propose an overview of the existing methods to quantify uncertainty associated with DL predictions. We focus on applications to medical image analysis, which present specific challenges due to the high dimensionality of images and their variable quality, as well as constraints associated with real-world clinical routine. Moreover, we discuss the concept of structural uncertainty, a corpus of methods to facilitate the alignment of segmentation uncertainty estimates with clinical attention. We then discuss the evaluation protocols to validate the relevance of uncertainty estimates. Finally, we highlight the open challenges for uncertainty quantification in the medical field.

Sparse annotation learning for dense volumetric MR image segmentation with uncertainty estimation

Objective.Training neural networks for pixel-wise or voxel-wise image segmentation is a challenging task that requires a considerable amount of training samples with highly accurate and densely delineated ground truth maps. This challenge becomes especially prominent in the medical imaging domain, where obtaining reliable annotations for training samples is a difficult, time-consuming, and expert-dependent process. Therefore, developing models that can perform well under the conditions of limited annotated training data is desirable.Approach.In this study, we propose an innovative framework called the extremely sparse annotation neural network (ESA-Net) that learns with only the single central slice label for 3D volumetric segmentation which explores both intra-slice pixel dependencies and inter-slice image correlations with uncertainty estimation. Specifically, ESA-Net consists of four specially designed distinct components: (1) an intra-slice pixel dependency-guided pseudo-label generation module that exploits uncertainty in network predictions while generating pseudo-labels for unlabeled slices with temporal ensembling; (2) an inter-slice image correlation-constrained pseudo-label propagation module which propagates labels from the labeled central slice to unlabeled slices by self-supervised registration with rotation ensembling; (3) a pseudo-label fusion module that fuses the two sets of generated pseudo-labels with voxel-wise uncertainty guidance; and (4) a final segmentation network optimization module to make final predictions with scoring-based label quantification.Main results.Extensive experimental validations have been performed on two popular yet challenging magnetic resonance image segmentation tasks and compared to five state-of-the-art methods.Significance.Results demonstrate that our proposed ESA-Net can consistently achieve better segmentation performances even under the extremely sparse annotation setting, highlighting its effectiveness in exploiting information from unlabeled data.

Application of uncertainty quantification to artificial intelligence in healthcare: A review of last decade (2013-2023)

Uncertainty estimation in healthcare involves quantifying and understanding the inherent uncertainty or variability associated with medical predictions, diagnoses, and treatment outcomes. In this era of Artificial Intelligence (AI) models, uncertainty estimation becomes vital to ensure safe decision-making in the medical field. Therefore, this review focuses on the application of uncertainty techniques to machine and deep learning models in healthcare. A systematic literature review was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Our analysis revealed that Bayesian methods were the predominant technique for uncertainty quantification in machine learning models, with Fuzzy systems being the second most used approach. Regarding deep learning models, Bayesian methods emerged as the most prevalent approach, finding application in nearly all aspects of medical imaging. Most of the studies reported in this paper focused on medical images, highlighting the prevalent application of uncertainty quantification techniques using deep learning models compared to machine learning models. Interestingly, we observed a scarcity of studies applying uncertainty quantification to physiological signals. Thus, future research on uncertainty quantification should prioritize investigating the application of these techniques to physiological signals. Overall, our review highlights the significance of integrating uncertainty techniques in healthcare applications of machine learning and deep learning models. This can provide valuable insights and practical solutions to manage uncertainty in real-world medical data, ultimately improving the accuracy and reliability of medical diagnoses and treatment recommendations.

Semi-Supervised Segmentation of Interstitial Lung Disease Patterns from CT Images via Self-Training with Selective Re-Training

Accurate segmentation of interstitial lung disease (ILD) patterns from computed tomography (CT) images is an essential prerequisite to treatment and follow-up. However, it is highly time-consuming for radiologists to pixel-by-pixel segment ILD patterns from CT scans with hundreds of slices. Consequently, it is hard to obtain large amounts of well-annotated data, which poses a huge challenge for data-driven deep learning-based methods. To alleviate this problem, we propose an end-to-end semi-supervised learning framework for the segmentation of ILD patterns (ESSegILD) from CT images via self-training with selective re-training. The proposed ESSegILD model is trained using a large CT dataset with slice-wise sparse annotations, i.e., only labeling a few slices in each CT volume with ILD patterns. Specifically, we adopt a popular semi-supervised framework, i.e., Mean-Teacher, that consists of a teacher model and a student model and uses consistency regularization to encourage consistent outputs from the two models under different perturbations. Furthermore, we propose introducing the latest self-training technique with a selective re-training strategy to select reliable pseudo-labels generated by the teacher model, which are used to expand training samples to promote the student model during iterative training. By leveraging consistency regularization and self-training with selective re-training, our proposed ESSegILD can effectively utilize unlabeled data from a partially annotated dataset to progressively improve the segmentation performance. Experiments are conducted on a dataset of 67 pneumonia patients with incomplete annotations containing over 11,000 CT images with eight different lung patterns of ILDs, with the results indicating that our proposed method is superior to the state-of-the-art methods.

Semi-Supervised 3D Medical Image Segmentation Based on Dual-Task Consistent Joint Learning and Task-Level Regularization

Semi-supervised learning has attracted wide attention from many researchers since its ability to utilize a few data with labels and relatively more data without labels to learn information. Some existing semi-supervised methods for medical image segmentation enforce the regularization of training by implicitly perturbing data or networks to perform the consistency. Most consistency regularization methods focus on data level or network structure level, and rarely of them focus on the task level. It may not directly lead to an improvement in task accuracy. To overcome the problem, this work proposes a semi-supervised dual-task consistent joint learning framework with task-level regularization for 3D medical image segmentation. Two branches are utilized to simultaneously predict the segmented and signed distance maps, and they can learn useful information from each other by constructing a consistency loss function between the two tasks. The segmentation branch learns rich information from both labeled and unlabeled data to strengthen the constraints on the geometric structure of the target. Experimental results on two benchmark datasets show that the proposed method can achieve better performance compared with other state-of-the-art works. It illustrates our method improves segmentation performance by utilizing unlabeled data and consistent regularization.

Semi-supervised OCT lesion segmentation via transformation-consistent with uncertainty and self-deep supervision

Optical coherence tomography (OCT) is a non-invasive, high-resolution ocular imaging technique with important implications for the diagnosis and management of retinal diseases. Automatic segmentation of lesions in OCT images is critical for assessing disease progression and treatment outcomes. However, existing methods for lesion segmentation require numerous pixel-wise annotations, which are difficult and time-consuming to obtain. To address this challenge, we propose a novel framework for semi-supervised OCT lesion segmentation, termed transformation-consistent with uncertainty and self-deep supervision (TCUS). To address the issue of lesion area blurring in OCT images and unreliable predictions from the teacher network for unlabeled images, an uncertainty-guided transformation-consistent strategy is proposed. Transformation-consistent is used to enhance the unsupervised regularization effect. The student network gradually learns from meaningful and reliable targets by utilizing the uncertainty information from the teacher network, to alleviate the performance degradation caused by potential errors in the teacher network's prediction results. Additionally, self-deep supervision is used to acquire multi-scale information from labeled and unlabeled OCT images, enabling accurate segmentation of lesions of various sizes and shapes. Self-deep supervision significantly improves the accuracy of lesion segmentation in terms of the Dice coefficient. Experimental results on two OCT datasets demonstrate that the proposed TCUS outperforms state-of-the-art semi-supervised segmentation methods.

Semi-supervised segmentation of coronary DSA using mixed networks and multi-strategies

The coronary arteries supply blood to the myocardium, which originate from the root of the aorta and mainly branch into the left and right. X-ray digital subtraction angiography (DSA) is a technique for evaluating coronary artery plaques and narrowing, that is widely used because of its time efficiency and cost-effectiveness. However, automated coronary vessel classification and segmentation remains challenging using a little data. Therefore, the purpose of this study is twofold: one is to propose a more robust method for vessel segmentation, the other is to provide a solution that is feasible with a small amount of labeled data. Currently, there are three main types of vessel segmentation methods, i.e., graphical- and statistical-based; clustering theory based, and deep learning-based methods for pixel-by-pixel probabilistic prediction, among which the last method is the mainstream with high accuracy and automation. Under this trend, an Inception-SwinUnet (ISUnet) network combining the convolutional neural network and Transformer basic module was proposed in this paper. Considering that data-driven fully supervised learning (FSL) segmentation methods require a large set of paired data with high-quality pixel-level annotation, which is expertise-demanding and time-consuming, we proposed a Semi-supervised Learning (SSL) method to achieve better performance with a small amount of labeled and unlabeled data. Different from the classical SSL method, i.e., Mean-Teacher, our method used two different networks for cross-teaching as the backbone. Meanwhile, inspired by deep supervision and confidence learning (CL), two effective strategies for SSL were adopted, which were denominated Pyramid-consistency Learning (PL) and Confidence Learning (CL), respectively. Both were designed to filter the noise and improve the credibility of pseudo labels generated by unlabeled data. Compared with existing methods, ours achieved superior segmentation performance over other FSL and SSL ones by using data with a small equal number of labels. Code is available in https://github.com/Allenem/SSL4DSA.

Residual attention based uncertainty-guided mean teacher model for semi-supervised breast masses segmentation in 2D ultrasonography

Breast tumor is the second deadliest disease among women around the world. Earlier tumor diagnosis is extremely important for improving the survival rate. Recent deep-learning techniques proved helpful in the timely diagnosis of various tumors. However, in the case of breast tumors, the characteristics of the tumors, i.e., low visual contrast, unclear boundary, and diversity in shape and size of breast lesions, make it more challenging to design a highly efficient detection system. Additionally, the scarcity of publicly available labeled data is also a major hurdle in the development of highly accurate and robust deep-learning models for breast tumor detection. To overcome these issues, we propose residual-attention-based uncertainty-guided mean teacher framework which incorporates the residual and attention blocks. The residual for optimizing the deep network by enabling the flow of high-level features and attention modules improves the focus of the model by optimizing its weights during the learning process. We further explore the potential of utilizing unlabeled data during the training process by employing the semi-supervised learning (SSL) method. Particularly, the uncertainty-guided mean-teacher student architecture is exploited to demonstrate the potential of incorporating the unlabeled samples during the training of residual attention U-Net model. The proposed SSL framework has been rigorously evaluated on two publicly available labeled datasets, i.e., BUSI and UDIAT datasets. The quantitative as well as qualitative results demonstrate that the proposed framework achieved competitive performance with respect to the previous state-of-the-art techniques and outperform the existing breast ultrasound masses segmentation techniques. Most importantly, the study demonstrates the potential of incorporating the additional unlabeled data for improving the performance of breast tumor segmentation.

Mass detection in automated 3-D breast ultrasound using a patch Bi-ConvLSTM network

Breast cancer mortality can be significantly reduced by early detection of its symptoms. The 3-D Automated Breast Ultrasound (ABUS) has been widely used for breast screening due to its high sensitivity and reproducibility. The large number of ABUS slices, and high variation in size and shape of the masses, make the manual evaluation a challenging and time-consuming process. To assist the radiologists, we propose a convolutional BiLSTM network to classify the slices based on the presence of a mass. Because of its patch-based architecture, this model produces the approximate location of masses as a heat map. The prepared dataset consists of 60 volumes belonging to 43 patients. The precision, recall, accuracy, F1-score, and AUC of the proposed model for slice classification were 84%, 84%, 93%, 84%, and 97%, respectively. Based on the FROC analysis, the proposed detector obtained a sensitivity of 82% with two false positives per volume.

Local contrastive loss with pseudo-label based self-training for semi-supervised medical image segmentation

Supervised deep learning-based methods yield accurate results for medical image segmentation. However, they require large labeled datasets for this, and obtaining them is a laborious task that requires clinical expertise. Semi/self-supervised learning-based approaches address this limitation by exploiting unlabeled data along with limited annotated data. Recent self-supervised learning methods use contrastive loss to learn good global level representations from unlabeled images and achieve high performance in classification tasks on popular natural image datasets like ImageNet. In pixel-level prediction tasks such as segmentation, it is crucial to also learn good local level representations along with global representations to achieve better accuracy. However, the impact of the existing local contrastive loss-based methods remains limited for learning good local representations because similar and dissimilar local regions are defined based on random augmentations and spatial proximity; not based on the semantic label of local regions due to lack of large-scale expert annotations in the semi/self-supervised setting. In this paper, we propose a local contrastive loss to learn good pixel level features useful for segmentation by exploiting semantic label information obtained from pseudo-labels of unlabeled images alongside limited annotated images with ground truth (GT) labels. In particular, we define the proposed contrastive loss to encourage similar representations for the pixels that have the same pseudo-label/GT label while being dissimilar to the representation of pixels with different pseudo-label/GT label in the dataset. We perform pseudo-label based self-training and train the network by jointly optimizing the proposed contrastive loss on both labeled and unlabeled sets and segmentation loss on only the limited labeled set. We evaluated the proposed approach on three public medical datasets of cardiac and prostate anatomies, and obtain high segmentation performance with a limited labeled set of one or two 3D volumes. Extensive comparisons with the state-of-the-art semi-supervised and data augmentation methods and concurrent contrastive learning methods demonstrate the substantial improvement achieved by the proposed method. The code is made publicly available at https://github.com/krishnabits001/pseudo_label_contrastive_training.

Uncertainty-aware deep learning in healthcare: A scoping review

Mistrust is a major barrier to implementing deep learning in healthcare settings. Entrustment could be earned by conveying model certainty, or the probability that a given model output is accurate, but the use of uncertainty estimation for deep learning entrustment is largely unexplored, and there is no consensus regarding optimal methods for quantifying uncertainty. Our purpose is to critically evaluate methods for quantifying uncertainty in deep learning for healthcare applications and propose a conceptual framework for specifying certainty of deep learning predictions. We searched Embase, MEDLINE, and PubMed databases for articles relevant to study objectives, complying with PRISMA guidelines, rated study quality using validated tools, and extracted data according to modified CHARMS criteria. Among 30 included studies, 24 described medical imaging applications. All imaging model architectures used convolutional neural networks or a variation thereof. The predominant method for quantifying uncertainty was Monte Carlo dropout, producing predictions from multiple networks for which different neurons have dropped out and measuring variance across the distribution of resulting predictions. Conformal prediction offered similar strong performance in estimating uncertainty, along with ease of interpretation and application not only to deep learning but also to other machine learning approaches. Among the six articles describing non-imaging applications, model architectures and uncertainty estimation methods were heterogeneous, but predictive performance was generally strong, and uncertainty estimation was effective in comparing modeling methods. Overall, the use of model learning curves to quantify epistemic uncertainty (attributable to model parameters) was sparse. Heterogeneity in reporting methods precluded the performance of a meta-analysis. Uncertainty estimation methods have the potential to identify rare but important misclassifications made by deep learning models and compare modeling methods, which could build patient and clinician trust in deep learning applications in healthcare. Efficient maturation of this field will require standardized guidelines for reporting performance and uncertainty metrics.

Graph-enhanced U-Net for semi-supervised segmentation of pancreas from abdomen CT scan

Objective. Accurate segmentation of the pancreas from abdomen CT scans is highly desired for diagnosis and treatment follow-up of pancreatic diseases. However, the task is challenged by large anatomical variations, low soft-tissue contrast, and the difficulty in acquiring a large set of annotated volumetric images for training. To overcome these problems, we propose a new segmentation network and a semi-supervised learning framework to alleviate the lack of annotated images and improve the accuracy of segmentation. Approach. In this paper, we propose a novel graph-enhanced pancreas segmentation network (GEPS-Net), and incorporate it into a semi-supervised learning framework based on iterative uncertainty-guided pseudo-label refinement. Our GEPS-Net plugs a graph enhancement module on top of the CNN-based U-Net to focus on the spatial relationship information. For semi-supervised learning, we introduce an iterative uncertainty-guided refinement process to update pseudo labels by removing low-quality and incorrect regions. Main results. Our method was evaluated by a public dataset with four-fold cross-validation and achieved the DC of 84.22%, improving 5.78% compared to the baseline. Further, the overall performance of our proposed method was the best compared with other semi-supervised methods trained with only 6 or 12 labeled volumes. Significance. The proposed method improved the segmentation performance of the pancreas in CT images under the semi-supervised setting. It will assist doctors in early screening and making accurate diagnoses as well as adaptive radiotherapy.

Breast Cancer Detection in Mammography Images Using Deep Convolutional Neural Networks and Fuzzy Ensemble Modeling Techniques

Breast cancer has evolved as the most lethal illness impacting women all over the globe. Breast cancer may be detected early, which reduces mortality and increases the chances of a full recovery. Researchers all around the world are working on breast cancer screening tools based on medical imaging. Deep learning approaches have piqued the attention of many in the medical imaging field due to their rapid growth. In this research, mammography pictures were utilized to detect breast cancer. We have used four mammography imaging datasets with a similar number of 1145 normal, benign, and malignant pictures using various deep CNN (Inception V4, ResNet-164, VGG-11, and DenseNet121) models as base classifiers. The proposed technique employs an ensemble approach in which the Gompertz function is used to build fuzzy rankings of the base classification techniques, and the decision scores of the base models are adaptively combined to construct final predictions. The proposed fuzzy ensemble techniques outperform each individual transfer learning methodology as well as multiple advanced ensemble strategies (Weighted Average, Sugeno Integral) with reference to prediction and accuracy. The suggested Inception V4 ensemble model with fuzzy rank based Gompertz function has a 99.32% accuracy rate. We believe that the suggested approach will be of tremendous value to healthcare practitioners in identifying breast cancer patients early on, perhaps leading to an immediate diagnosis.

DSGMFFN: Deepest semantically guided multi-scale feature fusion network for automated lesion segmentation in ABUS images

BACKGROUND AND OBJECTIVE

Automated breast ultrasound (ABUS) imaging technology has been widely used in clinical diagnosis. Accurate lesion segmentation in ABUS images is essential in computer-aided diagnosis (CAD) systems. Although deep learning-based approaches have been widely employed in medical image analysis, the large variety of lesions and the imaging interference make ABUS lesion segmentation challenging.

METHODS

In this paper, we propose a novel deepest semantically guided multi-scale feature fusion network (DSGMFFN) for lesion segmentation in 2D ABUS slices. In order to cope with the large variety of lesions, a deepest semantically guided decoder (DSGNet) and a multi-scale feature fusion model (MFFM) are designed, where the deepest semantics is fully utilized to guide the decoding and feature fusion. That is, the deepest information is given the highest weight in the feature fusion process, and participates in every decoding stage. Aiming at the challenge of imaging interference, a novel mixed attention mechanism is developed, integrating spatial self-attention and channel self-attention to obtain the correlation among pixels and channels to highlight the lesion region.

RESULTS

The proposed DSGMFFN is evaluated on 3742 slices of 170 ABUS volumes. The experimental result indicates that DSGMFFN achieves 84.54% and 73.24% in Dice similarity coefficient (DSC) and intersection over union (IoU), respectively.

CONCLUSIONS

The proposed method shows better performance than the state-of-the-art methods in ABUS lesion segmentation. Incorrect segmentation caused by lesion variety and imaging interference in ABUS images can be alleviated.

An Effective Semi-supervised Approach for Liver CT Image Segmentation

Despite the substantial progress made by deep networks in the field of medical image segmentation, they generally require sufficient pixel-level annotated data for training. The scale of training data remains to be the main bottleneck to obtain a better deep segmentation model. Semi-supervised learning is an effective approach that alleviates the dependence on labeled data. However, most existing semi-supervised image segmentation methods usually do not generate high-quality pseudo labels to expand training dataset. In this paper, we propose a deep semi-supervised approach for liver CT image segmentation by expanding pseudo-labeling algorithm under the very low annotated-data paradigm. Specifically, the output features of labeled images from the pretrained network combine with corresponding pixel-level annotations to produce class representations according to the mean operation. Then pseudo labels of unlabeled images are generated by calculating the distances between unlabeled feature vectors and each class representation. To further improve the quality of pseudo labels, we adopt a series of operations to optimize pseudo labels. A more accurate segmentation network is obtained by expanding the training dataset and adjusting the contributions between supervised and unsupervised loss. Besides, the novel random patch based on prior locations is introduced for unlabeled images in the training procedure. Extensive experiments show our method has achieved more competitive results compared with other semi-supervised methods when fewer labeled slices of LiTS dataset are available.

Shape and boundary-aware multi-branch model for semi-supervised medical image segmentation

Supervised learning-based medical image segmentation solutions usually require sufficient labeled training data. Insufficient available labeled training data often leads to the limitations of model performances, such as over-fitting, low accuracy, and poor generalization ability. However, this dilemma may worsen in the field of medical image analysis. Medical image annotation is usually labor-intensive and professional work. In this work, we propose a novel shape and boundary-aware deep learning model for medical image segmentation based on semi-supervised learning. The model makes good use of labeled data and also enables unlabeled data to be well applied by using task consistency loss. Firstly, we adopt V-Net for Pixel-wise Segmentation Map (PSM) prediction and Signed Distance Map (SDM) regression. In addition, we multiply multi-scale features, extracted by Pyramid Pooling Module (PPM) from input X, with 2 - |SDM| to enhance the features around the boundary of the segmented target, and then feed them into the Feature Fusion Module (FFM) for fine segmentation. Besides boundary loss, the high-level semantics implied in SDM facilitate the accurate segmentation of boundary regions. Finally, we get the ultimate result by fusing coarse and boundary-enhanced features. Last but not least, to mine unlabeled training data, we impose consistency constraints on the three core outputs of the model, namely PSM1, SDM, and PSM3. Through extensive experiments over three representative but challenging medical image datasets (LA2018, BraTS2019, and ISIC2018) and comparisons with the existing representative methods, we validate the practicability and superiority of our model.

Coarse-to-fine airway segmentation using multi information fusion network and CNN-based region growing

BACKGROUND AND OBJECTIVES

Automatic airway segmentation from chest computed tomography (CT) scans plays an important role in pulmonary disease diagnosis and computer-assisted therapy. However, low contrast at peripheral branches and complex tree-like structures remain as two mainly challenges for airway segmentation. Recent research has illustrated that deep learning methods perform well in segmentation tasks. Motivated by these works, a coarse-to-fine segmentation framework is proposed to obtain a complete airway tree.

METHODS

Our framework segments the overall airway and small branches via the multi-information fusion convolution neural network (Mif-CNN) and the CNN-based region growing, respectively. In Mif-CNN, atrous spatial pyramid pooling (ASPP) is integrated into a u-shaped network, and it can expend the receptive field and capture multi-scale information. Meanwhile, boundary and location information are incorporated into semantic information. These information are fused to help Mif-CNN utilize additional context knowledge and useful features. To improve the performance of the segmentation result, the CNN-based region growing method is designed to focus on obtaining small branches. A voxel classification network (VCN), which can entirely capture the rich information around each voxel, is applied to classify the voxels into airway and non-airway. In addition, a shape reconstruction method is used to refine the airway tree.

RESULTS

We evaluate our method on a private dataset and a public dataset from EXACT09. Compared with the segmentation results from other methods, our method demonstrated promising accuracy in complete airway tree segmentation. In the private dataset, the Dice similarity coefficient (DSC), Intersection over Union (IoU), false positive rate (FPR), and sensitivity are 93.5%, 87.8%, 0.015%, and 90.8%, respectively. In the public dataset, the DSC, IoU, FPR, and sensitivity are 95.8%, 91.9%, 0.053% and 96.6%, respectively.

CONCLUSION

The proposed Mif-CNN and CNN-based region growing method segment the airway tree accurately and efficiently in CT scans. Experimental results also demonstrate that the framework is ready for application in computer-aided diagnosis systems for lung disease and other related works.

Uncertainty-Guided Voxel-Level Supervised Contrastive Learning for Semi-Supervised Medical Image Segmentation

Semi-supervised learning reduces overfitting and facilitates medical image segmentation by regularizing the learning of limited well-annotated data with the knowledge provided by a large amount of unlabeled data. However, there are many misuses and underutilization of data in conventional semi-supervised methods. On the one hand, the model will deviate from the empirical distribution under the training of numerous unlabeled data. On the other hand, the model treats labeled and unlabeled data differently and does not consider inter-data information. In this paper, a semi-supervised method is proposed to exploit unlabeled data to further narrow the gap between the semi-supervised model and its fully-supervised counterpart. Specifically, the architecture of the proposed method is based on the mean-teacher framework, and the uncertainty estimation module is improved to impose constraints of consistency and guide the selection of feature representation vectors. Notably, a voxel-level supervised contrastive learning module is devised to establish a contrastive relationship between feature representation vectors, whether from labeled or unlabeled data. The supervised manner ensures that the network learns the correct knowledge, and the dense contrastive relationship further extracts information from unlabeled data. The above overcomes data misuse and underutilization in semi-supervised frameworks. Moreover, it favors the feature representation with intra-class compactness and inter-class separability and gains extra performance. Extensive experimental results on the left atrium dataset from Atrial Segmentation Challenge demonstrate that the proposed method has superior performance over the state-of-the-art methods.

Adaptive Hierarchical Dual Consistency for Semi-Supervised Left Atrium Segmentation on Cross-Domain Data

Semi-supervised learning provides great significance in left atrium (LA) segmentation model learning with insufficient labelled data. Generalising semi-supervised learning to cross-domain data is of high importance to further improve model robustness. However, the widely existing distribution difference and sample mismatch between different data domains hinder the generalisation of semi-supervised learning. In this study, we alleviate these problems by proposing an Adaptive Hierarchical Dual Consistency (AHDC) for the semi-supervised LA segmentation on cross-domain data. The AHDC mainly consists of a Bidirectional Adversarial Inference module (BAI) and a Hierarchical Dual Consistency learning module (HDC). The BAI overcomes the difference of distributions and the sample mismatch between two different domains. It mainly learns two mapping networks adversarially to obtain two matched domains through mutual adaptation. The HDC investigates a hierarchical dual learning paradigm for cross-domain semi-supervised segmentation based on the obtained matched domains. It mainly builds two dual-modelling networks for mining the complementary information in both intra-domain and inter-domain. For the intra-domain learning, a consistency constraint is applied to the dual-modelling targets to exploit the complementary modelling information. For the inter-domain learning, a consistency constraint is applied to the LAs modelled by two dual-modelling networks to exploit the complementary knowledge among different data domains. We demonstrated the performance of our proposed AHDC on four 3D late gadolinium enhancement cardiac MR (LGE-CMR) datasets from different centres and a 3D CT dataset. Compared to other state-of-the-art methods, our proposed AHDC achieved higher segmentation accuracy, which indicated its capability in the cross-domain semi-supervised LA segmentation.

A Multi-Task Learning Framework for Automated Segmentation and Classification of Breast Tumors From Ultrasound Images

Breast cancer is one of the most fatal diseases leading to the death of several women across the world. But early diagnosis of breast cancer can help to reduce the mortality rate. So an efficient multi-task learning approach is proposed in this work for the automatic segmentation and classification of breast tumors from ultrasound images. The proposed learning approach consists of an encoder, decoder, and bridge blocks for segmentation and a dense branch for the classification of tumors. For efficient classification, multi-scale features from different levels of the network are used. Experimental results show that the proposed approach is able to enhance the accuracy and recall of segmentation by 1.08%, 4.13%, and classification by 1.16%, 2.34%, respectively than the methods available in the literature.

Deep weakly-supervised breast tumor segmentation in ultrasound images with explicit anatomical constraints

Breast tumor segmentation is an important step in the diagnostic procedure of physicians and computer-aided diagnosis systems. We propose a two-step deep learning framework for breast tumor segmentation in breast ultrasound (BUS) images which requires only a few manual labels. The first step is breast anatomy decomposition handled by a semi-supervised semantic segmentation technique. The input BUS image is decomposed into four breast anatomical structures, namely fat, mammary gland, muscle and thorax layers. Fat and mammary gland layers are used as constrained region to reduce the search space for breast tumor segmentation. The second step is breast tumor segmentation performed in a weakly-supervised learning scenario where only image-level labels are available. Breast tumors are first recognized by a classification network and then segmented by the proposed class activation mapping and deep level set (CAM-DLS) method. For breast anatomy decomposition, the proposed framework achieves Dice similarity coefficient (DSC) of 83.0 ± 11.8%, 84.3 ± 10.0%, 80.7 ± 15.4% and 91.0 ± 11.4% for fat, mammary gland, muscle and thorax layers, respectively. For breast tumor recognition, the proposed framework achieves sensitivity of 95.8%, precision of 92.4%, specificity of 93.9%, accuracy of 94.8% and F1-score of 0.941. For breast tumor segmentation, the proposed framework achieves DSC of 77.3% and intersection-over-union (IoU) of 66.0%. In conclusion, the proposed framework could efficiently perform breast tumor recognition and segmentation simultaneously in a weakly-supervised setting with anatomical constraints.

High resolution histopathology image generation and segmentation through adversarial training

Semantic segmentation of histopathology images can be a vital aspect of computer-aided diagnosis, and deep learning models have been effectively applied to this task with varying levels of success. However, their impact has been limited due to the small size of fully annotated datasets. Data augmentation is one avenue to address this limitation. Generative Adversarial Networks (GANs) have shown promise in this respect, but previous work has focused mostly on classification tasks applied to MR and CT images, both of which have lower resolution and scale than histopathology images. There is limited research that applies GANs as a data augmentation approach for large-scale image semantic segmentation, which requires high-quality image-mask pairs. In this work, we propose a multi-scale conditional GAN for high-resolution, large-scale histopathology image generation and segmentation. Our model consists of a pyramid of GAN structures, each responsible for generating and segmenting images at a different scale. Using semantic masks, the generative component of our model is able to synthesize histopathology images that are visually realistic. We demonstrate that these synthesized images along with their masks can be used to boost segmentation performance, especially in the semi-supervised scenario.

Towards targeted ultrasound-guided prostate biopsy by incorporating model and label uncertainty in cancer detection

PURPOSE

Systematic prostate biopsy is widely used for cancer diagnosis. The procedure is blind to underlying prostate tissue micro-structure; hence, it can lead to a high rate of false negatives. Development of a machine-learning model that can reliably identify suspicious cancer regions is highly desirable. However, the models proposed to-date do not consider the uncertainty present in their output or the data to benefit clinical decision making for targeting biopsy.

METHODS

We propose a deep network for improved detection of prostate cancer in systematic biopsy considering both the label and model uncertainty. The architecture of our model is based on U-Net, trained with temporal enhanced ultrasound (TeUS) data. We estimate cancer detection uncertainty using test-time augmentation and test-time dropout. We then use uncertainty metrics to report the cancer probability for regions with high confidence to help the clinical decision making during the biopsy procedure.

RESULTS

Experiments for prostate cancer classification includes data from 183 prostate biopsy cores of 41 patients. We achieve an area under the curve, sensitivity, specificity and balanced accuracy of 0.79, 0.78, 0.71 and 0.75, respectively.

CONCLUSION

Our key contribution is to automatically estimate model and label uncertainty towards enabling targeted ultrasound-guided prostate biopsy. We anticipate that such information about uncertainty can decrease the number of unnecessary biopsy with a higher rate of cancer yield.

Dilated densely connected U-Net with uncertainty focus loss for 3D ABUS mass segmentation

BACKGROUND AND OBJECTIVE

Accurate segmentation of breast mass in 3D automated breast ultrasound (ABUS) images plays an important role in qualitative and quantitative ABUS image analysis. Yet this task is challenging due to the low signal to noise ratio and serious artifacts in ABUS images, the large shape and size variation of breast masses, as well as the small training dataset compared with natural images. The purpose of this study is to address these difficulties by designing a dilated densely connected U-Net (D²U-Net) together with an uncertainty focus loss.

METHODS

A lightweight yet effective densely connected segmentation network is constructed to extensively explore feature representations in the small ABUS dataset. In order to deal with the high variation in shape and size of breast masses, a set of hybrid dilated convolutions is integrated into the dense blocks of the D²U-Net. We further suggest an uncertainty focus loss to put more attention on unreliable network predictions, especially the ambiguous mass boundaries caused by low signal to noise ratio and artifacts. Our segmentation algorithm is evaluated on an ABUS dataset of 170 volumes from 107 patients. Ablation analysis and comparison with existing methods are conduct to verify the effectiveness of the proposed method.

RESULTS

Experiment results demonstrate that the proposed algorithm outperforms existing methods on 3D ABUS mass segmentation tasks, with Dice similarity coefficient, Jaccard index and 95% Hausdorff distance of 69.02%, 56.61% and 4.92 mm, respectively.

CONCLUSIONS

The proposed method is effective in segmenting breast masses on our small ABUS dataset, especially breast masses with large shape and size variations.

Cross-Model Attention-Guided Tumor Segmentation for 3D Automated Breast Ultrasound (ABUS) Images

Tumor segmentation in 3D automated breast ultrasound (ABUS) plays an important role in breast disease diagnosis and surgical planning. However, automatic segmentation of tumors in 3D ABUS images is still challenging, due to the large tumor shape and size variations, and uncertain tumor locations among patients. In this paper, we develop a novel cross-model attention-guided tumor segmentation network with a hybrid loss for 3D ABUS images. Specifically, we incorporate the tumor location into a segmentation network by combining an improved 3D Mask R-CNN head into V-Net as an end-to-end architecture. Furthermore, we introduce a cross-model attention mechanism that is able to aggregate the segmentation probability map from the improved 3D Mask R-CNN to each feature extraction level in the V-Net. Then, we design a hybrid loss to balance the contribution of each part in the proposed cross-model segmentation network. We conduct extensive experiments on 170 3D ABUS from 107 patients. Experimental results show that our method outperforms other state-of-the-art methods, by achieving the Dice similarity coefficient (DSC) of 64.57%, Jaccard coefficient (JC) of 53.39%, recall (REC) of 64.43%, precision (PRE) of 74.51%, 95th Hausdorff distance (95HD) of 11.91mm, and average surface distance (ASD) of 4.63mm. Our code will be available online (https://github.com/zhouyuegithub/CMVNet).