Ultrasonic Medical Tissue Imaging Using Probabilistic Inversion: Leveraging Variational Inference for Speed Reconstruction and Uncertainty Quantification

Abstract: Full Waveform Inversion (FWI) is a promising technique for achieving high-resolution imaging in medical ultrasound. Traditional FWI methods suffer from issues related to computational efficiency, dependence on initial models, and the inability to quantify uncertainty. This study introduces the Stein Variational Gradient Descent (SVGD) algorithm into FWI, aiming to improve inversion performance and enhance uncertainty quantification. By deriving the posterior gradient, the study explores the integration of SVGD with FWI and demonstrates its ability to approximate complex priors. In-silico experiments with synthetic data and real-world breast tissue data highlight the advantages of the SVGD-based framework over conventional FWI. SVGD-based FWI improves inversion quality, provides more reliable uncertainty quantification, and offers a tighter bound for the prior distribution. These findings show that probabilistic inversion is a promising tool for addressing the limitations of traditional FWI methods in ultrasonic imaging of medical tissues.