Improved 3D Reconstruction Algorithm for Ultrasound B-scan Image with Freehand Tracker

EM algorithm for the reconstruction of freehand B-Scan ultrasound image was developed by Joao M. Sanches et al. The reconstruction has a parameter. which can be adjusted so that the results can be smoother or sharper depending to the value of.. In order to make the image smoother inside the organs but sharper in their boundaries simultaneously, we introduced a improved EM algorithm: EM algorithm with a diffusion filer or is referred as EMD algorithm. There was a cubic average filter inside the loop of the iteration of the EM algorithm. This average filter is replaced by a diffusion filter in the EMD algorithm. The diffusion filter offers an additional parameter K(d) which can be used to adjust the reconstructed image with better optimization in both smoothness insider the human organ and sharpness in its boundary. Two above mentioned reconstruction algorithms for the freehand B-scan ultrasound image are compared through the simulation and the phantom measurements. In the simulation, strong noises are added to the ultrasound frame data. The parameters of two algorithms are optimized to get smallest errors. The errors are compared between two algorithms with optimized parameters. For the measurement with phantom, the Eigen's tracker system is used to continuously measure the coordinates of the ultrasound probe. The ultrasound B-scan frame is synchronously recorded with the probe coordinates. Zonare ultrasound machine is used to acquire the 2D frame images. The segmentation of the reconstruction results is done. The segmentation volumes of the prostate phantoms are compared. The results shows that EMD algorithm is better at reducing the noises and keeping the image edge comparing to EM algorithm. Eigen's tracker can acquire freehand ultrasound data for a 3D image reconstruction with high quality.