Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Ultrasound Image Enhancement with the Variance of Diffusion Models

Yuxin Zhang, Clément Huneau, Jérôme Idier, Diana Mateus

TL;DR

A novel approach that integrates adaptive beamforming with denoising diffusion-based variance imaging to address this challenge of high-quality ultrasound images by applying Eigenspace-Based Minimum Variance (EBMV) beamforming and employing a denoising diffusion model fine-tuned on ultrasound data.

Abstract

Ultrasound imaging, despite its widespread use in medicine, often suffers from various sources of noise and artifacts that impact the signal-to-noise ratio and overall image quality. Enhancing ultrasound images requires a delicate balance between contrast, resolution, and speckle preservation. This paper introduces a novel approach that integrates adaptive beamforming with denoising diffusion-based variance imaging to address this challenge. By applying Eigenspace-Based Minimum Variance (EBMV) beamforming and employing a denoising diffusion model fine-tuned on ultrasound data, our method computes the variance across multiple diffusion-denoised samples to produce high-quality despeckled images. This approach leverages both the inherent multiplicative noise of ultrasound and the stochastic nature of diffusion models. Experimental results on a publicly available dataset demonstrate the effectiveness of our method in achieving superior image reconstructions from single plane-wave acquisitions. The code is available at: https://github.com/Yuxin-Zhang-Jasmine/IUS2024_Diffusion.

Ultrasound Image Enhancement with the Variance of Diffusion Models

TL;DR

A novel approach that integrates adaptive beamforming with denoising diffusion-based variance imaging to address this challenge of high-quality ultrasound images by applying Eigenspace-Based Minimum Variance (EBMV) beamforming and employing a denoising diffusion model fine-tuned on ultrasound data.

Abstract

Ultrasound imaging, despite its widespread use in medicine, often suffers from various sources of noise and artifacts that impact the signal-to-noise ratio and overall image quality. Enhancing ultrasound images requires a delicate balance between contrast, resolution, and speckle preservation. This paper introduces a novel approach that integrates adaptive beamforming with denoising diffusion-based variance imaging to address this challenge. By applying Eigenspace-Based Minimum Variance (EBMV) beamforming and employing a denoising diffusion model fine-tuned on ultrasound data, our method computes the variance across multiple diffusion-denoised samples to produce high-quality despeckled images. This approach leverages both the inherent multiplicative noise of ultrasound and the stochastic nature of diffusion models. Experimental results on a publicly available dataset demonstrate the effectiveness of our method in achieving superior image reconstructions from single plane-wave acquisitions. The code is available at: https://github.com/Yuxin-Zhang-Jasmine/IUS2024_Diffusion.
Paper Structure (6 sections, 7 equations, 5 figures)

This paper contains 6 sections, 7 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Method
  3. Adaptive Beamforming
  4. Denoising Diffusion Variance Imaging
  5. Experimental validation on real data
  6. Conclusion

Figures (5)

  • Figure 1: Method overview
  • Figure 2: Comparison of reconstructed images on the PICMUS-EC dataset. The colored boundaries outline the regions where the evaluation metrics are calculated.
  • Figure 3: Statistical behavior of EBMV+DUS compared to EBMV on the PICMUS-EC dataset. The values at the position indicated by the dotted green line in the top-left image are compared in the right green plots. It shows that the variance of EBMV+DUS in the zero region enclosed by the dashed red box is non-zero. All grayscale images are in decibels with a dynamic range [-60,0].
  • Figure 4: Quantitative comparison of the PICMUS-EC dataset. A and L denote axial and lateral directions, respectively.
  • Figure 5: Comparison of reconstructed images on the PICMUS-CC dataset.