Speckle Tracking of Tendon Displacement in the Carpal Tunnel: Improved Quantification Using Singular Value Decomposition.

Ultrasound is a real-time image modality enabling the analysis of tendon dynamics for the diagnosis of carpal tunnel syndrome. Automatic tendon displacement quantification algorithms based on speckle tracking generally suffer from underestimation due to stationary background present in the tendon region. We propose an improved quantification method using singular value decomposition (SVD) filtering to suppress the clutter. The accuracy of our improved speckle tracking (IST) method was validated against a ground truth and compared to the accuracy of our original block matching (OBM) algorithm and commercial tissue tracking (CTT) software. The methods were evaluated in experiments involving six human cadaver arms. The ground-truth displacements were generated by tracking metal markers inserted in the tendons. The relative displacement errors with respect to the ground truth for IST were 12 ± 16.9%, which was significantly lower than for OBM (19.7 ± 20.8%) and for CTT (25.8 ± 18.4%). These findings show that SVD filtering improves the tendon tracking by reducing underestimation due to clutter.