Ultrasound Image Enhancement with the Variance of Diffusion Models

  • 2024-09-17 18:29:33
  • Yuxin Zhang, Clément Huneau, Jérôme Idier, Diana Mateus
  • 0

Abstract

Ultrasound imaging, despite its widespread use in medicine, often suffersfrom various sources of noise and artifacts that impact the signal-to-noiseratio and overall image quality. Enhancing ultrasound images requires adelicate balance between contrast, resolution, and speckle preservation. Thispaper introduces a novel approach that integrates adaptive beamforming withdenoising diffusion-based variance imaging to address this challenge. Byapplying Eigenspace-Based Minimum Variance (EBMV) beamforming and employing adenoising diffusion model fine-tuned on ultrasound data, our method computesthe variance across multiple diffusion-denoised samples to produce high-qualitydespeckled images. This approach leverages both the inherent multiplicativenoise of ultrasound and the stochastic nature of diffusion models. Experimentalresults on a publicly available dataset demonstrate the effectiveness of ourmethod in achieving superior image reconstructions from single plane-waveacquisitions. The code is available at:https://github.com/Yuxin-Zhang-Jasmine/IUS2024_Diffusion.

 

Quick Read (beta)

loading the full paper ...