SpeckleGAN: a generative adversarial network with an adaptive speckle layer to augment limited training data for ultrasound image processing

Lennart Bargsten,Alexander Schlaefer

doi:10.1007/s11548-020-02203-1

Abstract

PurposeIn the field of medical image analysis, deep learning methods gained huge attention over the last years. This can be explained by their often improved performance compared to classic explicit algorithms. In order to work well, they need large amounts of annotated data for supervised learning, but these are often not available in the case of medical image data. One way to overcome this limitation is to generate synthetic training data, e.g., by performing simulations to artificially augment the dataset. However, simulations require domain knowledge and are limited by the complexity of the underlying physical model. Another method to perform data augmentation is the generation of images by means of neural networks.MethodsWe developed a new algorithm for generation of synthetic medical images exhibiting speckle noise via generative adversarial networks (GANs). Key ingredient is a speckle layer, which can be incorporated into a neural network in order to add realistic and domain-dependent speckle. We call the resulting GAN architecture SpeckleGAN.ResultsWe compared our new approach to an equivalent GAN without speckle layer. SpeckleGAN was able to generate ultrasound images with very crisp speckle patterns in contrast to the baseline GAN, even for small datasets of 50 images. SpeckleGAN outperformed the baseline GAN by up to 165 % with respect to the Fréchet Inception distance. For artery layer and lumen segmentation, a performance improvement of up to 4 % was obtained for small datasets, when these were augmented with images by SpeckleGAN.ConclusionSpeckleGAN facilitates the generation of realistic synthetic ultrasound images to augment small training sets for deep learning based image processing. Its application is not restricted to ultrasound images but could be used for every imaging methodology that produces images with speckle such as optical coherence tomography or radar.

Highlights

Cardiovascular diseases like atherosclerosis are the leading cause of death globally [12]
The image sets generated by SpeckleGAN result in Fréchet Inception distance (FID) scores ranging from 134.0 for 50 training images to 113.4 for 435 training images
The quality of SpeckleGAN images only decreases slightly with fewer training examples, whereas the quality of images generated by the baseline generative adversarial networks (GANs) decreases strongly

Summary

Introduction

Cardiovascular diseases like atherosclerosis are the leading cause of death globally [12]. In recent years, finding diagnoses has been more and more supported by algorithms which provide additional information to the physician. Powerful deep learning methods gained significant importance due to their superior performance compared to many explicit algorithms. Annotations are usually made by trained experts to ensure high quality. This naturally leads to a lack of high-quality data. To overcome these limitations, data augmentation methods are commonly used [18]. In addition to applying random transformations to the data samples (which do not alter their labels), the generation of artificial training data is a possible way to enlarge

Methods

Results

Discussion

Conclusion