d Original Contribution HIGH-FRAME-RATE DEFORMATION IMAGING IN TWO DIMENSIONS USING CONTINUOUS SPECKLE-FEATURE TRACKING MARTIN V. ANDERSEN,* COOPER MOORE, y KRISTINE ARGES, z PETER SØGAARD, x LASSE R. ØSTERGAARD,* SAMUEL E. SCHMIDT ,* JOSEPH KISSLO, z and OLAF T. VON RAMM y *Department of Health Science and Technology, Aalborg University, Aalborg, Denmark; y Duke Biomedical Engineering Department, Duke University, Durham, North Carolina, USA; z Duke Department of Medicine, Duke University Hospital, Durham, North Carolina, USA; and x Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark (Received 26 October 2015; revised 8 July 2016; in final form 9 July 2016) Abstract—The study describes a novel algorithm for deriving myocardial strain from an entire cardiac cycle using high-frame-rate ultrasound images. Validation of the tracking algorithm was conducted in vitro prior to the appli- cation to patient images. High-frame-rate ultrasound images were acquired in vivo from 10 patients, and strain curves were derived in six myocardial regions around the left ventricle from the apical four-chamber view. Strain curves derived from high-frame-rate images had a higher frequency content than those derived using conventional methods, reflecting improved temporal sampling. (E-mail: mvan@hst.aau.dk) Ó 2016 World Federation for Ul- trasound in Medicine & Biology. Key Words: Deformation imaging, Strain, Algorithm, Speckle tracking, Feature, Speckle, Ultrasound, Echocardi- ology, High frame rate, Feature tracking. INTRODUCTION Cardiac disease is the leading cause of morbidity and mortality in the population over the age of 40. For example, heart failure is common in the United States, occurring in 6% to 10% of the adult population above the age of 65 (Roger et al. 2011). Heart failure can be the result of either electrical or mechanical abnormalities. The effects of both mechanical and electrical abnormal- ities are difficult to distinguish in early stages of heart failure, making accurate diagnosis and effective treat- ment difficult (Bristow et al. 2004; Risum et al. 2012, 2013; Vernooy et al. 2005). Detection and measurement of rapid mechanical phenomena, such as the propagation of mechanical contraction and relaxation subsequent to depolarization, could have great clinical significance. Depolarization propagates with a velocity between 0.5 and 2 m/s depend- ing on whether the electrical excitation is through the myocardial tissue or the Purkinje fibers (Durrer et al. 1970). It is unlikely that these events will be detected us- ing conventional ultrasound scanners, which operate at 50 to 110 fps (Teske et al. 2007). A frame rate of $500 fps is necessary to record information about mechanical activa- tion sequences and correlate them to electrical measure- ments acquired by electrocardiography (EKG) (Cikes et al. 2014). If these mechanical phenomena are similar in speed and duration to the propagation of the electrical excitation wave through the left ventricle, they will last 30 ms (Durrer et al. 1970). These rapid mechanical phenomena are currently undetected because of the low image acqui- sition rates of conventional clinical ultrasound imaging (Teske et al. 2007). A common method for automating measurements in ultrasound images is speckle or feature tracking. The most commonly used commercial speckle tracking algorithms, such as GE Healthcare’s Automated Function Imaging software and TomTec’s 2-D Cardiac Performance Anal- ysis software, use Optical Flow methods for frame-to- frame myocardial motion tracking (Geyer et al. 2010; Hor et al. 2011; Teske et al. 2007). Optical Flow has a lower limit of detectable velocities that depends on the algorithm’s ability to detect sub-pixel variations between frames. If the movement of a target between two sequen- tial images is smaller than that between adjoining spatial samples, and the tracking method does not implement sub-pixel variation detection, the frame-to-frame Optical Ultrasound in Med. & Biol., Vol. 42, No. 11, pp. 2606–2615, 2016 Copyright Ó 2016 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/$ - see front matter http://dx.doi.org/10.1016/j.ultrasmedbio.2016.07.009 Address correspondence to: Martin V. Andersen, Department of Health Science and Technology, Aalborg University, Fredrik Bajers Vej 7, 9220 Aalborg Ø, Denmark. E-mail: mvan@hst.aau.dk 2606