Search papers, labs, and topics across Lattice.
The paper introduces IRSDE-Despeckle, a diffusion-based ultrasound despeckling method leveraging the Image Restoration Stochastic Differential Equations (IRSDE) framework. The method is trained on a large, paired dataset of simulated ultrasound images generated from speckle-free MR images using the Matlab UltraSound Toolbox. Results demonstrate superior performance compared to classical filters and learning-based baselines on a held-out simulated test set, with uncertainty quantification showing correlation between prediction variance and reconstruction error.
A physics-grounded diffusion model trained on simulated data achieves state-of-the-art ultrasound despeckling, offering a practical route to improved image quality and diagnostic accuracy.
Ultrasound imaging is widely used for real-time, noninvasive diagnosis, but speckle and related artifacts reduce image quality and can hinder interpretation. We present a diffusion-based ultrasound despeckling method built on the Image Restoration Stochastic Differential Equations framework. To enable supervised training, we curate large paired datasets by simulating ultrasound images from speckle-free magnetic resonance images using the Matlab UltraSound Toolbox. The proposed model reconstructs speckle-suppressed images while preserving anatomically meaningful edges and contrast. On a held-out simulated test set, our approach consistently outperforms classical filters and recent learning-based despeckling baselines. We quantify prediction uncertainty via cross-model variance and show that higher uncertainty correlates with higher reconstruction error, providing a practical indicator of difficult or failure-prone regions. Finally, we evaluate sensitivity to simulation probe settings and observe domain shift, motivating diversified training and adaptation for robust clinical deployment.
