Estimation de la trajectoire d'une sonde ultrasonore pour l'échographie 3D main-libre sans capteur de position

Abstract

Freehand 3D ultrasound consists in acquiring a sequence of ultrasound images using a traditional 2D ultrasound probe, and combining them to reconstruct a 3D volume. This volumetric reconstruction step requires to register or obtain the orientations and positions of all the images relative to the first image in the sequence. Registration in sensorless freehand 3D ultrasound uses speckle decorrelation to estimate small rigid motions between pairs of 2D images. Trajectory estimation combines these motion measurements to obtain the position of each image relative to the first. This is prone to the accumulation of measurement bias. Whereas previous work concentrated on correcting biases at the source, this paper proposes to reduce error accumulation by carefully choosing the set of measurements used to estimate the trajectory. To do so, a graph is created with frames as vertices and motion measurements as edge. Weights represent measurement quality, which are estimated with Gaussian Process regression. By searching for constrained shortest paths in the graph, many trajectories are generated and averaged to obtain a more accurate final trajectory estimate. Results on speckle phantom imagery show significantly improved trajectory estimates in comparison with the state-of-the-art, promising accurate volumetric reconstruction.

Date	1 Aug 2016
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Catherine Laporte (Supervisor) & Jean-Marc Lina (Co-supervisor)