Ultrasound Domain Adaptation Using Frequency Domain Analysis

  title={Ultrasound Domain Adaptation Using Frequency Domain Analysis},
  author={M. Sharifzadeh and Ali Kafaei Zad Tehrani and Habib Benali and Hassan Rivaz},
  journal={2021 IEEE International Ultrasonics Symposium (IUS)},
A common issue in exploiting simulated ultrasound data for training neural networks is the domain shift problem, where the trained models on synthetic data are not generalizable to clinical data. Recently, Fourier Domain Adaptation (FDA) has been proposed in the field of computer vision to tackle the domain shift problem by replacing the magnitude of the low-frequency spectrum of a synthetic sample (source) with a real sample (target). This method is attractive in ultrasound imaging given that… 

Figures and Tables from this paper


Accounting for Domain Shift in Neural Network Ultrasound Beamforming
This work proposes a novel domain adaptation scheme that leverages both simulated and unlabeled in vivo data to train an in vivo beamformer and demonstrates qualitative and quantitative improvements in in vivo image quality compared to conventional DAS as well as other advanced beamformers.
Training Deep Network Ultrasound Beamformers With Unlabeled In Vivo Data
This work proposes a novel domain adaptation (DA) scheme to correct for domain shift by incorporating unlabeled in vivo data during training, and uses cycle-consistent generative adversarial networks to map between simulated and in vivoData in both the input and ground truth target domains.
Transfer Learning for Domain Adaptation in MRI: Application in Brain Lesion Segmentation
This study trained a CNN on legacy MR images of brain and evaluated the performance of the domain-adapted network on the same task with images from a different domain, substantially outperforming a similar network trained on the the same set of examples from scratch.
Generalizing Deep Learning for Medical Image Segmentation to Unseen Domains via Deep Stacked Transformation
A deep stacked transformation approach for domain generalization that can be generalized to the design of highly robust deep segmentation models for clinical deployment and reaches the performance of state-of theart fully supervised models that are trained and tested on their source domains.
Automated Breast Ultrasound Lesions Detection Using Convolutional Neural Networks
This paper proposes the use of deep learning approaches for breast ultrasound lesion detection and investigates three different methods: a Patch-based LeNet, a U-Net, and a transfer learning approach with a pretrained FCN-AlexNet.
FDA: Fourier Domain Adaptation for Semantic Segmentation
A simple method for unsupervised domain adaptation, whereby the discrepancy between the source and target distributions is reduced by swapping the low-frequency spectrum of one with the other, which results indicate that even simple procedures can discount nuisance variability in the data that more sophisticated methods struggle to learn away.
MSDAN: Multi-Scale Self-Attention Unsupervised Domain Adaptation Network for Thyroid Ultrasound Images
A Multi-Scale Self-Attention Unsupervised Domain Adaptive framework (MSDAN), which learns the geometric information of the pathological tissues in medical images through the self-attention module, thereby improving the transfer effect of the semantic information inmedical images.
Paper presented at the 10th Nordic-Baltic Conference on Biomedical Imaging: Field: A Program for Simulating Ultrasound Systems
A program for the simulation of ultrasound systems is presented. It is based on the Tupholme-Stepanishen method, and is fast because of the use of a fareld approximation. Any kind of transducer
U-Net: Convolutional Networks for Biomedical Image Segmentation
It is shown that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks.
Domain Adaptive Faster R-CNN for Object Detection in the Wild
This work builds on the recent state-of-the-art Faster R-CNN model, and design two domain adaptation components, on image level and instance level, to reduce the domain discrepancy, based on $$-divergence theory.