Investigation of 3D Reconstruction Algorithms For Wireless Freehand Ultrasound Imaging System

The handheld 3D ultrasound imaging technique based on position tracking systems has been rapidly developed and widely applied in recent decades. The objectives of this study are to investigate the performance and accuracy of different 3D reconstruction algorithms including Voxel Nearest Neighbor (VNN), Pose Optimization Based (POB), and Implicit Representation (IR) methods. The high-precision phantom was used as the validation model to measure 2D/3D distance on the reconstructed image volume, and the measurements were evaluated with the true values obtained by caliber. The results indicated that the IR method presented the best reconstruction visualization and the smallest reconstruction errors for different motion cases. It demonstrated that the neural network-based reconstruction method can improve image quality and reduce reconstruction errors for the wireless freehand 3D ultrasound imaging systems.