P3A-1 Directional Fusion-based Multi-Volume Rendering Technique for 3D Color Doppler Imaging

We present a new multi-volume rendering technique (i.e., directional fusion) to integrate 3D anatomical structure from B-mode imaging with flow information from color Doppler imaging. In this technique, negative and positive flows are separated and independently rendered, and then these two rendered frames are merged with the rendered B-mode frame using alpha blending to generate a final 3D color Doppler image. Due to its independent rendering in separately handling negative and positive flows, this directional fusion method can preserve the directionality in the flow, unlike the conventional multi-volume rendering method where two directional flows are rendered together. To evaluate the feasibility of directional fusion, we have implemented the proposed rendering technique on a programmable ultrasound machine with four digital signal processors running at 450 MHz. For 128 times 128 times 128 B-mode and color Doppler volumes, the proposed method takes 141 ms, which corresponds to 7.1 volumes/s. In addition, we have evaluated the proposed directional fusion-based multi-volume technique on in vivo liver data acquired using a commercial ultrasound machine (i.e., EUB-8500, Hitachi Medical Corporation, Japan). The proposed fusion technique can clearly visualize the fused tissue structures and vascularity while preserving directionality in the flow. These preliminary results indicate that although it takes more computation due to the additional rendering pipeline, it can still be implemented in real time on modern ultrasound systems with 3D imaging capability while providing improved vascularity and tissue structure visualization