Monte Carlo Algorithm Accelerated Distance-Weighted Spatiotemporal Non-Local Means Filtering in Ultrafast Power Doppler Imaging

Ultrasound power Doppler (uPDI) imaging faces the challenge of high background noise due to unfocused plane wave transmit. Spatiotemporal non-local means filtering (ST-NLM) on clutter filtered blood flow data is an effective way to improve the uPDI quality, by suppressing noise and enhancing the small vessels simultaneously. There is room for improvement in the running speed of conventional ST-NLM. In this study, a Monte Carlo algorithm accelerated distance-weighted spatiotemporal non-local means filtering (MWST-NLM) method was proposed to achieve higher quality uPDI with high computational efficiency. In-vivo contrast-free human liver and in-vivo contrast-enhanced rat kidney data were used to validate the proposed method. Contrast-to-noise ratio (CNR), peak-to-side level (PSL), noise power and full width at half maximum (FWHM) were used for quantitative evaluation. For the rat kidney data, MWST-NLM achieves 9.47 dB higher CNR, 17.66 dB higher PSL, 6.52 dB lower noise power and 58.06 µm narrower FWHM than conventional ST-NLM. For the human liver data, MWST-NLM achieves 15.10 dB higher CNR, 16.89 dB higher PSL, 10.33 dB lower noise power and 186.53 µm narrower FWHM than conventional ST-NLM. In addition, the running time of the proposed method is more than 2 times faster than that of ST-NLM, demonstrating its potential for clinical applications.