Ultrasound Datasets Research Articles

Cross-image Dependency Modelling for Breast Ultrasound Segmentation.

We present a novel deep network (namely BUSSeg) equipped with both within- and cross-image long-range dependency modeling for automated lesions segmentation from breast ultrasound images, which is a quite daunting task due to (1) the large variation of breast lesions, (2) the ambiguous lesion boundaries, and (3) the existence of speckle noise and artifacts in ultrasound images. Our work is motivated by the fact that most existing methods only focus on modeling the within-image dependencies while neglecting the cross-image dependencies, which are essential for this task under limited training data and noise. We first propose a novel cross-image dependency module (CDM) with a cross-image contextual modeling scheme and a cross-image dependency loss (CDL) to capture more consistent feature expression and alleviate noise interference. Compared with existing cross-image methods, the proposed CDM has two merits. First, we utilize more complete spatial features instead of commonly used discrete pixel vectors to capture the semantic dependencies between images, mitigating the negative effects of speckle noise and making the acquired features more representative. Second, the proposed CDM includes both intra- and inter-class contextual modeling rather than just extracting homogeneous contextual dependencies. Furthermore, we develop a parallel bi-encoder architecture (PBA) to tame a Transformer and a convolutional neural network to enhance BUSSeg's capability in capturing within-image long-range dependencies and hence offer richer features for CDM. We conducted extensive experiments on two representative public breast ultrasound datasets, and the results demonstrate that the proposed BUSSeg consistently outperforms state-of-the-art approaches in most metrics.

Rethinking the unpretentious U-net for medical ultrasound image segmentation

Breast tumor segmentation from ultrasound images is one of the key steps that help us characterize and localize tumor regions. However, variable tumor morphology, blurred boundaries, and similar intensity distributions bring challenges for radiologists to segment breast tumors manually. During clinical diagnosis, there are higher demands on the segmentation accuracy and efficiency of breast ultrasound images, so there is an urgent need for an automated method to improve the segmentation accuracy as a technical tool to assist diagnosis. Inspired by the U-net and its many variations, this paper proposed an unpretentious nested U-net (NU-net) for accurate and efficient breast tumor segmentation. The key idea is to utilize U-nets with different depths and shared weights to achieve robust characterization of breast tumors. Specifically, we first utilize the deeper U-net (fifteen layers) as the backbone network to extract more sufficient breast tumor features. Then, we developed a multi-output U-net to be taken as the bond between the encoder and the decoder to enhance the network adaptability for breast tumors with different scales. Finally, the short-connection based on multi-step down-sampling is used to enhance the correlation of long-range information of encoded features. Extensive experimental results with fifteen state-of-the-art segmentation methods on three public breast ultrasound datasets demonstrate that our method has a more competitive segmentation performance on breast tumors. Furthermore, the robustness of our approach is further illustrated by the segmentation of renal ultrasound images. The source code is publicly available on https://github.com/CGPxy/NU-net.

BSMNet: Boundary-salience multi-branch network for intima-media identification in carotid ultrasound images

Carotid artery intima-media thickness (CIMT) is an essential factor in signaling the risk of cardiovascular diseases, which is commonly evaluated using ultrasound imaging. However, automatic intima-media segmentation and thickness measurement are still challenging due to the boundary ambiguity of intima-media and inherent speckle noises in ultrasound images. In this work, we propose an end-to-end boundary-salience multi-branch network, BSMNet, to tackle the carotid intima-media identification from ultrasound images, where the prior shape knowledge and anatomical dependence are exploited using a parallel linear structure learning modules followed by a boundary refinement module. Moreover, we design a strip attention model to boost the thin strip region segmentation with shape priors, in which an anisotropic kernel shape captures long-range global relations and scrutinizes meaningful local salient contexts simultaneously. Extensive experimental results on an in-house carotid ultrasound (US) dataset demonstrate the promising performance of our method, which achieves about 0.02 improvement in Dice and HD95 than other state-of-the-art methods. Our method is promising in advancing the analysis of systemic arterial disease with ultrasound imaging.

Weakly supervised real-time instance segmentation for ultrasound images of median nerves

Medical ultrasound technology has garnered significant attention in recent years, with Ultrasound-guided regional anesthesia (UGRA) and carpal tunnel diagnosis (CTS) being two notable examples. Instance segmentation, based on deep learning approaches, is a promising choice to support the analysis of ultrasound data. However, many instance segmentation models cannot achieve the requirement of ultrasound technology e.g. real-time. Moreover, fully supervised instance segmentation models require large numbers of images and corresponding mask annotations for training, which can be time-consuming and labor-intensive in the case of medical ultrasound data. This paper proposes a novel weakly supervised framework, CoarseInst, to achieve real-time instance segmentation of ultrasound images with only box annotations. CoarseInst not only improves the network structure, but also proposes a two-stage “coarse-to-fine” training strategy. Specifically, median nerves are used as the target application for UGRA and CTS. CoarseInst consists of two stages, with pseudo mask labels generated in the coarse mask generation stage for self-training. An object enhancement block is incorporated to mitigate the performance loss caused by parameter reduction in this stage. Additionally, we introduce a pair of loss functions, the amplification loss, and the deflation loss, that work together to generate the masks. A center area mask searching algorithm is also proposed to generate labels for the deflation loss. In the self-training stage, a novel self-feature similarity loss is designed to generate more precise masks. Experimental results on a practical ultrasound dataset demonstrate that CoarseInst could achieve better performance than some state-of-the-art fully supervised works.

Super-resolution of 2D ultrasound images and videos

This paper proposes a novel deep-learning framework for super-resolution ultrasound images and videos in terms of spatial resolution and line reconstruction. To this end, we up-sample the acquired low-resolution image through a vision-based interpolation method; then, we train a learning-based model to improve the quality of the up-sampling. We qualitatively and quantitatively test our model on different anatomical districts (e.g., cardiac, obstetric) images and with different up-sampling resolutions (i.e., 2X, 4X). Our method improves the PSNR median value with respect to SOTA methods of 1.7% on obstetric 2X raw images, 6.1% on cardiac 2X raw images, and 4.4% on abdominal raw 4X images; it also improves the number of pixels with a low prediction error of 9.0% on obstetric 4X raw images, 5.2% on cardiac 4X raw images, and 6.2% on abdominal 4X raw images. The proposed method is then applied to the spatial super-resolution of 2D videos, by optimising the sampling of lines acquired by the probe in terms of the acquisition frequency. Our method specialises trained networks to predict the high-resolution target through the design of the network architecture and the loss function, taking into account the anatomical district and the up-sampling factor and exploiting a large ultrasound data set. The use of deep learning on large data sets overcomes the limitations of vision-based algorithms that are general and do not encode the characteristics of the data. Furthermore, the data set can be enriched with images selected by medical experts to further specialise the individual networks. Through learning and high-performance computing, the proposed super-resolution is specialised to different anatomical districts by training multiple networks. Furthermore, the computational demand is shifted to centralised hardware resources with a real-time execution of the network’s prediction on local devices.Graphical abstract

Efficient Lung Ultrasound Classification.

A machine learning method for classifying lung ultrasound is proposed here to provide a point of care tool for supporting a safe, fast, and accurate diagnosis that can also be useful during a pandemic such as SARS-CoV-2. Given the advantages (e.g., safety, speed, portability, cost-effectiveness) provided by the ultrasound technology over other examinations (e.g., X-ray, computer tomography, magnetic resonance imaging), our method was validated on the largest public lung ultrasound dataset. Focusing on both accuracy and efficiency, our solution is based on an efficient adaptive ensembling of two EfficientNet-b0 models reaching 100% of accuracy, which, to our knowledge, outperforms the previous state-of-the-art models by at least 5%. The complexity is restrained by adopting specific design choices: ensembling with an adaptive combination layer, ensembling performed on the deep features, and minimal ensemble using two weak models only. In this way, the number of parameters has the same order of magnitude of a single EfficientNet-b0 and the computational cost (FLOPs) is reduced at least by 20%, doubled by parallelization. Moreover, a visual analysis of the saliency maps on sample images of all the classes of the dataset reveals where an inaccurate weak model focuses its attention versus an accurate one.

AAU-Net: An Adaptive Attention U-Net for Breast Lesions Segmentation in Ultrasound Images.

Various deep learning methods have been proposed to segment breast lesions from ultrasound images. However, similar intensity distributions, variable tumor morphologies and blurred boundaries present challenges for breast lesions segmentation, especially for malignant tumors with irregular shapes. Considering the complexity of ultrasound images, we develop an adaptive attention U-net (AAU-net) to segment breast lesions automatically and stably from ultrasound images. Specifically, we introduce a hybrid adaptive attention module (HAAM), which mainly consists of a channel self-attention block and a spatial self-attention block, to replace the traditional convolution operation. Compared with the conventional convolution operation, the design of the hybrid adaptive attention module can help us capture more features under different receptive fields. Different from existing attention mechanisms, the HAAM module can guide the network to adaptively select more robust representation in channel and space dimensions to cope with more complex breast lesions segmentation. Extensive experiments with several state-of-the-art deep learning segmentation methods on three public breast ultrasound datasets show that our method has better performance on breast lesions segmentation. Furthermore, robustness analysis and external experiments demonstrate that our proposed AAU-net has better generalization performance in the breast lesion segmentation. Moreover, the HAAM module can be flexibly applied to existing network frameworks. The source code is available on https://github.com/CGPxy/AAU-net.

A robust and automatic CT-3D ultrasound registration method based on segmentation, context, and edge hybrid metric.

The fusion of computed tomography (CT) and ultrasound (US) image can enhance lesion detection ability and improve the success rate of liver interventional radiology. The image-based fusion methods encounter the challenge of registration initialization due to the random scanning pose and limited field of view of US. Existing automatic methods those used vessel geometric information and intensity-based metric are sensitive to parameters and have low success rate. The learning-based methods require a large number of registered datasets for training. The aim of this study is to provide a fully automatic and robust US-3D CT registration method without registered training data and user-specified parameters assisted by the revolutionary deep learning-based segmentation, which can further be used for preparing training samples for the study of learning-basedmethods. We propose a fully automatic CT-3D US registration method by two improved registration metrics. We propose to use 3D U-Net-based multi-organ segmentation of US and CT to assist the conventional registration. The rigid transform is searched in the space of any paired vessel bifurcation planes where the best transform is decided by a segmentation overlap metric, which is more related to the segmentation precision than Dice coefficient. In nonrigid registration phase, we propose a hybrid context and edge based image similarity metric with a simple mask that can remove most noisy US voxels to guide the B-spline transform registration. We evaluate our method on 42 paired CT-3D US datasets scanned with two different US devices from two hospitals. We compared our methods with other exsiting methods with both quantitative measures of target registration error (TRE) and the Jacobian determinent with paired t-test and qualitative registration imagingresults. The results show that our method achieves fully automatic rigid registration TRE of 4.895 mm, deformable registration TRE of 2.995 mm in average, which outperforms state-of-the-art automatic linear methods and nonlinear registration metrics with paired t-test's p value less than 0.05. The proposed overlap metric achieves better results than self similarity description (SSD), edge matching (EM), and block matching (BM) with p values of 1.624E-10, 4.235E-9, and 0.002, respectively. The proposed hybrid edge and context-based metric outperforms context-only, edge-only, and intensity statistics-only-based metrics with p values of 0.023, 3.81E-5, and 1.38E-15, respectively. The 3D US segmentation has achieved mean Dice similarity coefficient (DSC) of 0.799, 0.724, 0.788, and precision of 0.871, 0.769, 0.862 for gallbladder, vessel, and branch vessel,respectively. The deep learning-based US segmentation can achieve satisfied result to assist robust conventional rigid registration. The Dice similarity coefficient-based metrics, hybrid context, and edge image similarity metric contribute to robust and accurateregistration.

Immunohistochemical Index Prediction of Breast Cancer by Using Hybrid Ultrasound Data.

The immunohistochemical index is significant to help the selection of treatment strategy for breast cancer patients. Existing studies that focus on conventional ultrasound features and certain types of immunohistochemistry expressions are limited to correlation exploration, and only few studies have built predictive models. In this study, a Tri-Branch deep learning network is built for prediction of the immunohistochemical HER2 using the hybrid ultrasound data, instead of relying on the invasive and biopsy-based histopathological examination. Specifically, the deep learning model uses the cross-model attention and the interactive learning approaches to obtain the strong complementarity of hybrid data comprising B-mode US, contrast-enhanced ultrasound, and optical flow motion information to enhance accuracy of immunohistochemical HER2 prediction. The proposed prediction model was evaluated using hybrid ultrasound dataset from 335 breast cancer patients. The experimental results indicated that the Tri-Branch model had a high accuracy of 86.23% for HER2 status prediction, and the HER2 status prediction for patients with different pathology grades exhibited some meaningful clinical observations.

DSEU-net: A novel deep supervision SEU-net for medical ultrasound image segmentation

The automatic and accurate medical ultrasound image segmentation has been a challenging task due to the coupled interference of various internal and external factors. In recent years, CNN techniques have been widely and successfully used in medical image segmentation. Motivated by this, this paper proposes a novel squeeze-and-excitation attention U-net with deep supervision (DSEU-net) for medical ultrasound image segmentation. Specifically, a deeper U-net is first used as a benchmark network to capture sufficient target feature information from complex ultrasound images. Then, the squeeze-and-excitation (SE) block is regarded as the bond between encoder and decoder to enhance the attention to useful object regions. Moreover, the introduction of SE block not only strengthens the association of useful information at a distance, but also suppresses the introduction of irrelevant information. Finally, the deep supervised constraints are added to the decoding stage of the network to refine the prediction masks of ultrasound images. Extensive experimental results on three clinical ultrasound datasets show that DSEU-net has better robustness and superiority in ultrasound image segmentation. In the segmentation of the first breast ultrasound dataset (BUSI), the values of Jaccard, Precision, Recall, Specificity and Dice are 70.36%, 79.73%, 82.70%, 97.42% and 78.51%, respectively. The values of Jaccard, Precision, Recall, Specificity and Dice for our method on the second breast ultrasound dataset (Dataset B) are 73.17%, 82.58%, 84.02%, 99.05% and 81.50%, respectively. For the segmentation of kidney ultrasound dataset (KUS), the values of Jaccard, Precision, Recall, Specificity, Dice, HD, ASSD and ABD are 89.47, 94.77, 94.36, 99.10, 94.32, 12.42, 0.48 and 3.44, respectively. Comparing with the original U-net, DSEU-net improved on average 8.28% and 12.55% on five metrics for two breast ultrasound data. DSEU-net improved on average 54.81% on eight metrics for the kidney ultrasound dataset.

Recent Advances in Artificial Intelligence-Assisted Ultrasound Scanning

Ultrasound (US) is a flexible imaging modality used globally as a first-line medical exam procedure in many different clinical cases. It benefits from the continued evolution of ultrasonic technologies and a well-established US-based digital health system. Nevertheless, its diagnostic performance still presents challenges due to the inherent characteristics of US imaging, such as manual operation and significant operator dependence. Artificial intelligence (AI) has proven to recognize complicated scan patterns and provide quantitative assessments for imaging data. Therefore, AI technology has the potential to help physicians get more accurate and repeatable outcomes in the US. In this article, we review the recent advances in AI-assisted US scanning. We have identified the main areas where AI is being used to facilitate US scanning, such as standard plane recognition and organ identification, the extraction of standard clinical planes from 3D US volumes, and the scanning guidance of US acquisitions performed by humans or robots. In general, the lack of standardization and reference datasets in this field makes it difficult to perform comparative studies among the different proposed methods. More open-access repositories of large US datasets with detailed information about the acquisition are needed to facilitate the development of this very active research field, which is expected to have a very positive impact on US imaging.

Robust and Realtime Large Deformation Ultrasound Registration Using End-to-End Differentiable Displacement Optimisation

Image registration for temporal ultrasound sequences can be very beneficial for image-guided diagnostics and interventions. Cooperative human-machine systems that enable seamless assistance for both inexperienced and expert users during ultrasound examinations rely on robust, realtime motion estimation. Yet rapid and irregular motion patterns, varying image contrast and domain shifts in imaging devices pose a severe challenge to conventional realtime registration approaches. While learning-based registration networks have the promise of abstracting relevant features and delivering very fast inference times, they come at the potential risk of limited generalisation and robustness for unseen data; in particular, when trained with limited supervision. In this work, we demonstrate that these issues can be overcome by using end-to-end differentiable displacement optimisation. Our method involves a trainable feature backbone, a correlation layer that evaluates a large range of displacement options simultaneously and a differentiable regularisation module that ensures smooth and plausible deformation. In extensive experiments on public and private ultrasound datasets with very sparse ground truth annotation the method showed better generalisation abilities and overall accuracy than a VoxelMorph network with the same feature backbone, while being two times faster at inference.

Ultrasound Image Segmentation and Classification of Benign and Malignant Thyroid Nodules on the Basis of Deep Learning

This study aimed to investigate the effect of an image denoising algorithm based on weighted low-rank matrix restoration on thyroid nodule ultrasound images. A total of 1000 original ultrasound image data sets of thyroid nodules were selected as the study samples. The nodule segmentation data set of thyroid ultrasound region of interest (ROI) images was drawn and acquired. By introducing multiscale features and an attention mechanism to optimize the U-Net model, an ultrasound image segmentation model (F-U-Net) was constructed. The performance of the traditional U network model and full convolutional neural network model (FCN) was analyzed and compared by simulation experiments. The results showed that the dice coefficient, accuracy, and recall of the improved loss function in this study were significantly higher than those of the traditional cross entropy loss function and dice coefficient loss function, and the differences were statistically significant (P < 0.05). The Dice coefficient, accuracy, and recall of the F-U-net model were significantly higher than those of the traditional FCN model and U-net model (P < 0.05). The diagnostic sensitivity, specificity, accuracy, and positive predictive value of the F-U-net model for benign and malignant thyroid nodules were significantly higher than those of the FCN model and U-net model (P < 0.05). In summary, the proposed F-U network can effectively process the ultrasound images of thyroid nodules, improve the image quality, and help to improve the diagnostic effect of benign and malignant thyroid nodules. It provides a data reference for segmentation and reconstruction of benign and malignant ultrasound images of thyroid nodules.

HCTNet: A hybrid CNN-transformer network for breast ultrasound image segmentation

Automatic breast ultrasound image segmentation helps radiologists to improve the accuracy of breast cancer diagnosis. In recent years, the convolutional neural networks (CNNs) have achieved great success in medical image analysis. However, it exhibits limitations in modeling long-range relations, which is unfavorable for ultrasound images with speckle noise and shadows, resulting in decreased accuracy of breast lesion segmentation. Transformer can obtain sufficient global information, but it is deficient in acquiring local details and needs to be pre-trained on large-scale datasets. In this paper, we propose a Hybrid CNN-Transformer network (HCTNet) for boosting the breast lesion segmentation in ultrasound images. In the encoder of HCTNet, Transformer Encoder Blocks (TEBlocks) are designed to learn the global contextual information, which are combined with CNNs to extract features. In the decoder of HCTNet, a Spatial-wise Cross Attention (SCA) module is developed based on the spatial attention mechanism, which reduces the semantic discrepancy with the encoder. Moreover, residual connection is used between decoder blocks to make the generated features more discriminative by aggregating contextual feature maps at different semantic scales. Extensive experiments on three public breast ultrasound datasets demonstrate that HCTNet outperforms other medical image segmentation methods and the recent semantic segmentation methods on breast ultrasound lesion segmentation.

Method for Carotid Artery 3-D Ultrasound Image Segmentation Based on CSWin Transformer.

Precise segmentation of carotid artery (CA) structure is an important prerequisite for the medical assessment and detection of carotid plaques. For automatic segmentation of the media-adventitia boundary (MAB) and lumen-intima boundary (LIB) in 3-D ultrasound images of the CA, a U-shaped CSWin transformer (U-CSWT) is proposed. Both the encoder and decoder of the U-CSWT are composed of hierarchical CSWT modules, which can capture rich global context information in the 3-D image. Experiments were performed on a 3-D ultrasound image data set of the CA, and the results indicate that the U-CSWT performs better than other convolutional neural network (CNN)-based and CNN-transformer hybrid methods. The model yields Dice coefficients of 94.6 ± 3.0% and 90.8 ± 5.1% for the MAB and LIB in the common carotid artery (CCA) and 92.9 ± 4.9% and 89.6 ± 6.2% for MAB and LIB in the bifurcation, respectively. Our U-CSWT is expected to become an effective method for automatic segmentation of 3-D ultrasound images of CA.

Ultrasonic phased array inspection of defects in additive manufactured aluminum alloy considering velocity anisotropy

The detection of defect in additive manufactured (AM) metal materials is of great importance for industrial applications, and remains a challenge due to the heterogeneity and anisotropic of AM materials. This study presents an internal defects detection method for WAAM aluminum alloy using phased-array ultrasound. The detailed ultrasonic testing guidance considering the properties of WAAM metal material for inspecting and locating the internal defects is provided. Phased array ultrasound multi-angle-scanning pattern is applied for detecting the AM parts more reliably. The multi-angle data fusion algorithm is applied to manage the raw data obtained by different angle beams for generating ultrasound data sets encoded with physical information. The velocity of ultrasound beams with different angles propagated in the WAAM material is investigated and the relationship between the velocity and the angle of incidence can be established using the quadratic regression analysis. A modified data fusion algorithm considering the actual velocity variations is proposed. Comparisons are made between the actual depth of the artificial defect in WAAM deposited 2319 aluminum alloy block and depth measured by data fusion algorithm without/with velocity anisotropy correction. Results show the modified algorithm improves the accuracy of defect localization.

ATFE-Net: Axial Transformer and Feature Enhancement-based CNN for ultrasound breast mass segmentation

Breast mass is one of the main clinical symptoms of breast cancer. Recently, many CNN-based methods for breast mass segmentation have been proposed. However, these methods have difficulties in capturing long-range dependencies, causing poor segmentation of large-scale breast masses. In this paper, we propose an axial Transformer and feature enhancement-based CNN (ATFE-Net) for ultrasound breast mass segmentation. Specially, an axial Transformer (Axial-Trans) module and a Transformer-based feature enhancement (Trans-FE) module are proposed to capture long-range dependencies. Axial-Trans module only calculates self-attention in width and height directions of input feature maps, which reduces the complexity of self-attention significantly from O(n2) to O(n). In addition, Trans-FE module can enhance feature representation by capturing dependencies between different feature layers, since deeper feature layers have richer semantic information and shallower feature layers have more detailed information. The experimental results show that our ATFE-Net achieved better performance than several state-of-the-art methods on two publicly available breast ultrasound datasets, with Dice coefficient of 82.46% for BUSI and 86.78% for UDIAT, respectively.

Liver Tumors Segmentation Using 3D SegNet Deep Learning Approach

An ultrasonic filter detects signs of malignant tumors by analysing the image’s pixel quality fluctuations caused by a liver ailment. Signs of malignant growth proximity are identified in an ultrasound filter through image pixel quality variations from a liver’s condition. Those changes are more common in alcoholic liver conditions than in other etiologies of cirrhosis, suggesting that the cause may be alcohol instead of liver disease. Existing Two-Dimensional (2D) ultrasound data sets contain an accuracy rate of 85.9% and a 2D Computed Tomography (CT) data set of 91.02%. The most recent work on designing a Three-Dimensional (3D) ultrasound imaging system in or close to real-time is examined. In this article, a Deep Learning (DL) model is implemented and modified to fit liver CT segmentation, and a semantic pixel classification of road scenes is recommended. The architecture is called semantic pixel-wise segmentation and comprises a hierarchical link of encoder-decoder layers. A standard data set was used to test the proposed model for liver CT scans and the tumor accuracy in the training phase. For the normal class, we obtained 100% precision for chronic cirrhosis hepatitis (73%), offset cirrhosis (59.26%), and offensive cirrhosis (91.67%) for chronic hepatitis or cirrhosis (73,0%). The aim is to develop a Computer-Aided Detection (CAD) screening tool to detect steatosis. The results proved 98.33% exactness, 94.59% sensitivity, and 92.11% case with Convolutional Neural Networks (CNN) classification. Although the classifier’s performance did not differentiate so clearly at this level, it was recommended that CNN generally perform better due to the good relationship between Area under the Receiver Operating Characteristics Curve (AUC) and accuracy.

Speckle Noise Suppression in Ultrasound Images Using Modular Neural Networks

In spite of the advancement in computerized imaging, many image modalities produce images with commotion influencing both the visual quality and upsetting quantitative image analysis. In this way, the research in the zone of image denoising is very dynamic. Among an extraordinary assortment of image restoration and denoising techniques the neural network system-based noise suppression is a basic and productive methodology. In this paper, Bilateral Filter (BF) based Modular Neural Networks (MNN) has been utilized for speckle noise suppression in the ultrasound image. Initial step the BF filter is used to filter the input image. From the output of BF, statistical features such as mean, standard deviation, median and kurtosis have been extracted and these features are used to train the MNN. Then, the filtered images from the BF are again denoised using MNN. The ultrasound dataset from the Kaggle site is used for the training and testing process. The simulation outcomes demonstrate that the BF-MNN filtering method performs better for the multiplicative noise concealment in UltraSound (US) images. From the simulation results, it has been observed that BF-MNN performs better than the existing techniques in terms of peak signal to noise ratio (34.89), Structural Similarity Index (0.89) and Edge Preservation Index (0.67).

Anatomy-Aware Contrastive Representation Learning for Fetal Ultrasound.

Self-supervised contrastive representation learning offers the advantage of learning meaningful visual representations from unlabeled medical datasets for transfer learning. However, applying current contrastive learning approaches to medical data without considering its domain-specific anatomical characteristics may lead to visual representations that are inconsistent in appearance and semantics. In this paper, we propose to improve visual representations of medical images via anatomy-aware contrastive learning (AWCL), which incorporates anatomy information to augment the positive/negative pair sampling in a contrastive learning manner. The proposed approach is demonstrated for automated fetal ultrasound imaging tasks, enabling the positive pairs from the same or different ultrasound scans that are anatomically similar to be pulled together and thus improving the representation learning. We empirically investigate the effect of inclusion of anatomy information with coarse- and fine-grained granularity, for contrastive learning and find that learning with fine-grained anatomy information which preserves intra-class difference is more effective than its counterpart. We also analyze the impact of anatomy ratio on our AWCL framework and find that using more distinct but anatomically similar samples to compose positive pairs results in better quality representations. Extensive experiments on a large-scale fetal ultrasound dataset demonstrate that our approach is effective for learning representations that transfer well to three clinical downstream tasks, and achieves superior performance compared to ImageNet supervised and the current state-of-the-art contrastive learning methods. In particular, AWCL outperforms ImageNet supervised method by 13.8% and state-of-the-art contrastive-based method by 7.1% on a cross-domain segmentation task. The code is available at https://github.com/JianboJiao/AWCL.