Volumetric Visualization of the Dermal Vasculature with Signal and Image-based Feature Extraction on a High-frequency Ultrafast Ultrasound Dataset.

Abstract

Objective: To effectively observe the complex vascular network of the skin, 3-D volumetric imaging was performed via continuous scanning of a 1-D high-frequency ultrasound array transducer. In comparison to step-wise scanning of volumetric imaging, the acquisition of ultrasonic signal with continuous scanning is faster yet synchronous with a trade-off of transducer translation noise, which impacts the filtering of flow signal and image contrast. In this study, we proposed a processing algorithm to overcome this noise and enhance the visibility of the dermal vasculature.

Methods: The proposed approach involved acquiring volumetric data from the dorsum of the hand. The acquired data were segmented along the elevational direction into sections, and region-based singular value decomposition (SVD) filtering was applied to each section to extract the flow signal. Later two-step de-noising was performed based on non-local means (NLM) and top-hat (TH) morphological filters to effectively remove background noise caused by the continuous scanning. All of the processed sections were reconstructed and rendered in a volumetric viewer for 3-D representation of the vasculature.

Results: Section-wise, region-based SVD processing highlighted the presence of the vasculature in each 2-D slice. The implementation of NLM and TH filters removed the background scanning noise and significantly enhanced image quality (contrast-to-noise ratio) through 12.87 ± 1.3 dB (before NLM) to 18.91 ± 2.3 dB (after NLM) and 62.61 ± 19.2 dB (after NLM + TH).

Conclusion: Comprehensive visualization of the dermal vasculature in a 3-D volumetric view revealed the feasibility and significance of the proposed approach. This provides an insight into the visualization of complex vasculatures in 3-D with a 1-D array transducer, thereby enhancing diagnostic techniques in skin imaging.