Common motion estimator accuracy in shear wave-based optical coherence elastography
Justin R. Rippy 1, Manmohan Singh 1, Salavat Aglyamov 2, and Kirill V. Larin 1,3
1 Department of Biomedical Engineering, University of Houston, 4800 Calhoun Rd. Houston, TX, 77204, USA
2 Department of Mechanical Engineering, University of Houston, 4726 Calhoun Rd., Houston, TX 77204, USA
3 Molecular Physiology and Biophysics, Baylor College of Medicine, One Baylor Plaza, BCM335, Houston, TX 77030, USA
Abstract
Motion estimators are often used in shear wave optical coherence elastography (OCE) to measure small displacements when estimating group velocity. Three commonly used motion estimators, Kasai, Loupas, and the vector method, were compared with one another across several amplitudes and SNRs. We use data collected from shear wave OCE on 8%, 12%, and 16% gelatin phantoms to create simulation data to test the effects of amplitude and SNR on group velocity measurement. We generate normalized noiseless profiles via linear regression before adjusting the amplitudes to the desired values and adding in white Gaussian noise. Then, we reconstruct the profiles using the motion estimators and calculate group velocities by cross correlation. We show that the vector method is better at both reconstruction and group velocity calculation in low-amplitude, low-SNR situations. There is a lower bound on SNR at approximately 20 dB that is necessary for accurate group velocity measurement for any of the methods. After an SNR of 50 dB, the percent error is less than 5% no matter the amplitude or motion estimator, which is useful when trying to balance other parameters that might lower SNR. Additionally, the vector method has a more extensive coverage of lower error values than either Kasai or Loupas methods. We show that the vector method is superior for most cases, though it hasn’t been extensively used for dynamic elastography.
File with abstract
Speaker
Justin R. Rippy
University of Houston
United States
Report
File with report
Discussion
Ask question
