Joint image registration and volume reconstruction for 3D ultrasound

摘要

Three-dimensional (3D) ultrasound aims to reconstruct a 3D volume of data from a set of ultrasound images. To achieve this goal, accurate measurements of the probe position and orientation are required. This is not always possible in practice, due to the limitation of current spatial location systems. Unfortunately, position errors produce a misalignment of the volume cross-sections which blur the reconstruction results. To overcome this difficulty an alignment algorithm must be derived. This paper describes a method which performs both tasks: volume reconstruction and image alignment. The proposed algorithm is based on the optimization of a single objective function which depends on both sets of parameters. Reconstruction results with synthetic and medical data are presented showing significant improvement compared to the results obtained with standard technique without alignment.