Self-supervised Deep Learning for Denoising in Ultrasound Microvascular Imaging

Ultrasound microvascular imaging (UMI) has emerged as a powerful, noninvasive modality for visualizing the microvasculature, offering valuable insights into underlying pathological and physiological processes. However, it is often hindered by low signal-to-noise ratio (SNR) of unfocused plane wave imaging, especially in contrast-free or deep tissue scenarios, which may impair subsequent vascular quantification and reliable disease diagnosis. To address this challenge, we propose Half-Angle-to-Half-Angle (HA2HA), a self-supervised denoising framework specifically designed for UMI. HA2HA constructs training pairs from complementary angular subsets of beamformed radio-frequency (RF) blood flow data, across which vascular signals remain consistent while noise varies. HA2HA was trained using in-vivo contrast-free pig kidney data and evaluated on independent test datasets from unseen subjects, including contrast-free and contrast-enhanced pig kidney, human kidney and human liver data. Compared to the conventional method, HA2HA achieved a maximum contrast-to-noise ratio (CNR) improvement of approximately 10 dB in the contrast-enhanced pig kidney data, indicating a substantial enhancement in image quality. In addition to power Doppler imaging, denoising directly in the RF domain is also beneficial for other downstream processes, such as color Doppler imaging (CDI). CDI results of human liver derived from the HA2HA-denoised signals exhibited improved microvascular flow visualization, with a suppressed noisy background. HA2HA offers a label-free and generalizable solution for robust vascular imaging in both contrast-free and contrast-enhanced UMI.