Enhanced Shear Wave Velocity Calculation in Glaucoma Patients using Adaptive Singular Value Filter and Deep Neural Network in Ultrasound Elastography

Objective: The purpose of this study was to develop an adaptive singular value decomposition filter (SVF) method for improving calculation of shear wave velocity (SWV) on two locations of cornea and sclera for patients with glaucoma. Methods: Three patients with glaucoma and three healthy controls were enrolled in this study. We employed two independents adaptive SVFs for two regions of interest (ROI) of cornea and sclera (SVF-ROI). The deep neural network (DNN) was used to select the high efficiency cutoff value for SVF-ROI of two ROIs of cornea and sclera for reducing the noise of B-mode imaging and SWV. Results: We observed a trend of improved SNR when considering SWV values using three methods (i.e., no filter, the SVF, and SVF-ROI) for both eyes at the sclera (13.9, 14.8, and 19.6, respectively) and cornea (4.32, 4.29, and 9.78, respectively). Conclusion: significant improvement in SNR of SWV were found between the eyes with glaucoma and healthy eyes. Significance: The SVF-ROI provides an innovative method to increase accuracy of the vibro-ultrasound technique for assessing patients with glaucoma.