Learning a tissue invariant ultrasound speckle decorrelation model
ABSTRACT In untracked freehand 3D ultrasound (US), image content can be used to infer the trajectory of the transducer without a position tracking device. The nominal relationship between image correlation and elevational separation is established from controlled scans of a speckle phantom and used to determine out-of-plane motion. Unfortunately, this nominal relationship only holds under Rayleigh scattering conditions, which rarely occur in real tissue. This paper presents a method for learning the elevational correlation length of US signals in arbitrary tissue from a set of example synthetic US scans using sparse Gaussian process regression. Experiments on synthetic and real imagery of animal tissue show that the data driven approach generalises well across transducers, yielding results of accuracy superior to a base-line speckle detection approach and comparable to the state of the art. Additionally, the new approach uniquely provides a measure of uncertainty in the estimated correlation length.
SourceAvailable from: Tal Arbel[Show abstract] [Hide abstract]
ABSTRACT: In freehand 3D ultrasound, out-of-plane transducer motion can be estimated via speckle decorrelation instead of using a position tracking device. This approach was recently adapted to arbitrary media by predicting elevational decorrelation curves from local image statistics. However, such adaptive models tend to yield biased measurements in the presence of spatially persistent structures. To account for such failures, this paper introduces a new iterative algorithm for probabilistic fusion and selection of correlation measurements. In experiments with imagery of animal tissue, the approach yields significant accuracy improvements over alternatives which do not apply principled measurement selection.
[Show abstract] [Hide abstract]
ABSTRACT: In freehand 3D ultrasound (US), the relative positions and orientations of the 2D US images are usually obtained from a position tracking device, at the expense of clinical convenience. As an alternative or complement to this approach, transducer motion can be inferred from image content, using image registration techniques to recover in-plane motion and speckle decorrelation to recover out-of-plane motion. One difficulty with the speckle decorrelation approach is that for real tissues, the rate of speckle decorrelation is not only transducer dependent, but also medium dependent. This paper proposes a novel method for estimating the elevational correlation length of US signals in such media by learning its relationship to in-plane image statistics from a pool of synthetic US imagery generated from virtual phantoms of varied micro-structure. Learning takes place within a sparse Gaussian process regression framework. In experiments with synthetic US imagery and real imagery of animal tissue, the approach is shown to generalise well across transducer and medium changes, with performance better than a method based on speckle classification and comparable to our implementation of the heuristic state-of-the-art method. The proposed approach better lends itself to improvement through the creation of more realistic training sets.Medical image analysis 04/2011; 15(2):202-13. DOI:10.1016/j.media.2010.08.006 · 3.68 Impact Factor