Variational Formulation of Unsupervised Deep Learning for Ultrasound Image Artifact Removal

Khan, Shujaat; Huh, Jaeyoung; Ye, Jong Chul

doi:10.1109/tuffc.2021.3056197

Cited by 24 publications

(22 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…proposed an unsupervised deep learning approach based on an optimal transport‐driven CycleGAN network for ultrasound image artifact removal. In their model, an unsupervised learning problem was formulated as a stochastic generalization of the penalized least squares using an optimal transport theory 96 . Unsupervised CNNs have been proposed to predict time‐delay estimation between RF data in ultrasound elastography.…”

Section: Deep Learning Technologiesmentioning

confidence: 99%

“…This approach is particularly useful in finding meaningful patterns and groups in the dataset, and for extracting generative features. Unsupervised learning approaches in ultrasound imaging have been used for tasks such as image artifact removal, 96 and image registration and segmentation. 97 Khan et al 96 proposed an unsupervised deep learning approach based on an optimal transport-driven CycleGAN network for ultrasound image artifact removal.…”

Section: Advanced Learning Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Deep learning in ultrasound elastography imaging: A review

Bhatt

et al. 2022

Medical Physics

View full text Add to dashboard Cite

It is known that changes in the mechanical properties of tissues are associated with the onset and progression of certain diseases. Ultrasound elastography is a technique to characterize tissue stiffness using ultrasound imaging either by measuring tissue strain using quasi‐static elastography or natural organ pulsation elastography, or by tracing a propagated shear wave induced by a source or a natural vibration using dynamic elastography. In recent years, deep learning has begun to emerge in ultrasound elastography research. In this review, several common deep learning frameworks in the computer vision community, such as multilayered perceptron, convolutional neural network, and recurrent neural network, are described. Then, recent advances in ultrasound elastography using such deep learning techniques are revisited in terms of algorithm development and clinical diagnosis. Finally, the current challenges and future developments of deep learning in ultrasound elastography are prospected.

show abstract

Section: Deep Learning Technologiesmentioning

confidence: 99%

Section: Advanced Learning Modelsmentioning

confidence: 99%

Deep learning in ultrasound elastography imaging: A review

Bhatt

et al. 2022

Medical Physics

View full text Add to dashboard Cite

show abstract

“…From the 2D visualization of original feature space X , the decoded-output using proposed DeepLSE X = Dec(Enc(X )), and the residual of original-input and decoded output of DeepLSE X − X in Figure 7a it can be seen that the original feature space X with 1D-PCA-GCNR score of 0.14 units, is highly overlapping/non-linear and have almost no discriminating capabilities in linear-space. Whereas, the proposed DeepLSE method which acts as optimaltransport (OT) [35,36], and shift the data distributions in such a way that the reconstruction error of the original signal remains comparable to the conventional AE as well as the classification power is greatly improved. Its feature space, shown in Figure 7b, results in 1D-PCA-GCNR score of 0.77 units, which is 0.63 units higher than the original discriminating power.…”

Section: Analysis Of Decoder and Residual Errormentioning

confidence: 99%

AoP-LSE: Antioxidant Proteins Classification Using Deep Latent Space Encoding of Sequence Features

Usman

Khan

Park

et al. 2021

CIMB

Self Cite

View full text Add to dashboard Cite

It is of utmost importance to develop a computational method for accurate prediction of antioxidants, as they play a vital role in the prevention of several diseases caused by oxidative stress. In this correspondence, we present an effective computational methodology based on the notion of deep latent space encoding. A deep neural network classifier fused with an auto-encoder learns class labels in a pruned latent space. This strategy has eliminated the need to separately develop classifier and the feature selection model, allowing the standalone model to effectively harness discriminating feature space and perform improved predictions. A thorough analytical study has been presented alongwith the PCA/tSNE visualization and PCA-GCNR scores to show the discriminating power of the proposed method. The proposed method showed a high MCC value of 0.43 and a balanced accuracy of 76.2%, which is superior to the existing models. The model has been evaluated on an independent dataset during which it outperformed the contemporary methods by correctly identifying the novel proteins with an accuracy of 95%.

show abstract

“…The authors show that by doing this the quality of generated Images can be improved considerably. Other works such as [16] where the authors used cycleGAN to model the super resolution in ultrasound imaging by artifact removal in an unsupervised fashion. The authors [16] also mention that this work is able to tackle other similar objective works involving deconvolution, speckle removal amongst many others.…”

Section: Introductionmentioning

confidence: 99%

“…Other works such as [16] where the authors used cycleGAN to model the super resolution in ultrasound imaging by artifact removal in an unsupervised fashion. The authors [16] also mention that this work is able to tackle other similar objective works involving deconvolution, speckle removal amongst many others. There have also been self-supervised learning based approaches for super resolution keeping in mind the lack of available high resolution images in [17] and thus prove to be superior to state of the art super resolution algorithms for Ultrasound image super resolution on some datasets.…”

Section: Introductionmentioning

confidence: 99%

Patch Based Transformation for Minimum Variance Beamformer Image Approximation Using Delay and Sum Pipeline

Sairoop¹,

Madhavanunni²,

Panicker³

2021

Preprint

View full text Add to dashboard Cite

In the recent past, there have been several efforts in accelerating computationally heavy beamforming algorithms such as minimum variance distortionless response (MVDR) beamforming to achieve real-time performance comparable to the popular delay and sum (DAS) beamforming. This has been achieved using a variety of neural network architectures ranging from fully connected neural networks (FCNNs), convolutional neural networks (CNNs) and general adversarial networks (GANs). However most of these approaches are working with optimizations considering image level losses and hence require a significant amount of dataset to ensure that the process of beamforming is learned. In this work, a patch level U-Net based neural network is proposed, where the delay compensated radio frequency (RF) patch for a fixed region in space (e.g. 32x32) is transformed through a U-Net architecture and multiplied with DAS apodization weights and optimized for similarity with MVDR image of the patch. Instead of framing the beamforming problem as a regression problem to estimate the apodization weights, the proposed approach treats the non-linear transformation of the RF data space that can account for the data driven weight adaptation done by the MVDR approach in the parameters of the network. In this way, it is also observed that by restricting the input to a patch the model will learn the beamforming pipeline as an image non-linear transformation problem.

show abstract

Variational Formulation of Unsupervised Deep Learning for Ultrasound Image Artifact Removal

Cited by 24 publications

References 42 publications

Deep learning in ultrasound elastography imaging: A review

Deep learning in ultrasound elastography imaging: A review

AoP-LSE: Antioxidant Proteins Classification Using Deep Latent Space Encoding of Sequence Features

Patch Based Transformation for Minimum Variance Beamformer Image Approximation Using Delay and Sum Pipeline

Contact Info

Product

Resources

About