Official Proceedings - AIUM

American Institute of Ultrasound in Medicine Proceedings J Ultrasound Med 32(suppl):S1–S134, 2013as c bfvaried from 1350 to 1550 m/s and from 1440 to 1640 m/s whenscanning phantoms F and N-F, respectively, in steps of 10 m/s. The c bfatwhich S cwas minimized was chosen as the c testimate. The c testimatebias was computed as the interplane average of the percentage differencefrom substitution measurements (F phantom, 1452 m/s; N-F phantom,1544 m/s).Results—The minimum required ROI size was 20 pulselengths and 14 uncorrelated scan lines (7 × 9 mm 2 in this experiment). Ingeneral, ct was underestimated by –1.0% ± 0.1% and by –1.3% ± 0.3% forthe F and N-F phantoms, respectively. No significant differences werefound among estimates from different excitation frequencies or amongthose at the physical location or the nominal location of the transmit focus.Conclusions—Results indicate that estimates can be performedwithin 1 cm 2 at the location of the nominal transmit focus, whichcan be a fixed parameter during the in vivo application of this method.The method is being applied in the characterization of liver diseases andlesions.Cardiovascular UltrasoundModerator: John Blebea, MD, MBA1521708 Lagrangian Deformation Tracking of the Left Ventricle forCardiac Ultrasound Strain ImagingChi Ma,* Tomy Varghese Medical Physics, University ofWisconsin, Madison, Wisconsin USAObjectives—Lagrangian description of myocardial tissue structuredeformation is key to accurate regional strain estimation of the leftventricular wall over time. Failure to couple the estimated displacementand strain information with the correct myocardial tissue structures willlead to erroneous result in the displacement and strain distribution overtime.Methods—This study presents a method to obtain Lagrangianbaseddisplacement tracking. Myocardial issue structures are divided intoa fixed number of pixels whose deformation is tracked over the cardiaccycle. An algorithm that uses a polar grid generated between the estimatedendocardial and epicardial contours for the left ventricle in cardiac shortaxisimages is proposed to ensure Lagrangian description of the pixels.Displacement estimations from consecutive radiofrequency frames werethen mapped onto the polar grid to obtain a distribution of the actual displacementthat is mapped to the polar grid over time.Results—The method was validated against a finite-element–based canine heart model coupled with an ultrasound simulation program.Segmental analysis of the accumulated displacement and strain over a cardiaccycle demonstrated excellent agreement between the ideal result obtaineddirectly from the finite-element model and our Lagrangian approachto strain estimation. Traditional Eulerian-based estimation results, on theother hand, showed a significant deviation from the ideal result. An in vivocomparison of the displacement and strain estimated using parasternalshort-axis views is also presented.Conclusions—Lagrangian strain estimation using a polar griddemonstrates accurate results when validated in a finite-element cardiacmodel. In addition to the cardiac application, this approach can also beapplied to transverse scans of arteries, where a polar grid can be generatedbetween the contours delineating the outer and inner walls of the vesselsfrom the blood flowing though the vessels. (Supported by National Institutesof Health grants 5R21EB010098 and R01CA112192-S103.)S441509371 Effects of Respiration on Estimation of Systolic PulmonaryArtery Pressure in Patients With Right Ventricle SystolicDysfunctionXiao-Yong Zhang,* Tie-Sheng Cao, Li-Jun Yuan UltrasoundDiagnostics, Tangdu Hospital, Fourth Military MedicalUniversity, Xi’an, ChinaObjectives—We investigated the effects of respiration on thepeak velocity of tricuspid regurgitation (TR) and estimation of systolicpulmonary artery pressure (SPAP) in patients with right ventricle (RV)systolic dysfunction by Doppler echocardiography.Methods—Continuous wave Doppler spectra of TR wererecorded in 32 patients with and 28 controls without RV systolic dysfunction.Electrocardiography and respiratory tracing were recorded simultaneously.The expiratory and inspiratory peak velocities of TR wereacquired and averaged for 5 consecutive respiratory cycles. The SPAPduring expiration and inspiration was calculated.Results—The velocity of TR and SPAP did not vary significantlybetween expiration and inspiration in controls (2.77 ± 0.23 and 2.82 ± 0.26m/s; P = .776; 35.94 ± 4.96 and 36.18 ± 5.12 mm Hg; P = .747), whereasthe velocity of TR and SPAP decreased significantly from expiration to inspirationin patients with RV systolic dysfunction (3.27 ± 0.35 and 2.59 ±0.22 m/s; P < .001; 53.72 ± 7.39 and 38.45 ± 5.63 mm Hg; P < .001).Conclusions—Respiration has significant effects on the velocityof TR and SPAP in patients with RV systolic dysfunction, and the measurementshould be carried out when patients are at the end of expiration.1541517 High–Frame Rate Lateral Strain Estimation Using VirtualBeam Focusing in Canine and Human Hearts In VivoEthan Bunting, 1 * Jean Provost, 1 Elisa Konofagou 1,21Biomedical Engineering, 2 Radiology, Columbia University,New York, New York USAObjectives—Ultrasonic strain imaging is capable of providing clinicianswith useful information regarding cardiac function in a fast, noninvasivemanner. Lateral strain estimation is required to obtain the full 2D straintensor of the heart and accurately represent the cardiac deformation within theimage plane. Previous work has shown that 2 major parameters influencingthe quality of lateral strain estimation are the motion estimation rate and beamdensity (Provost et al. Phys Med Biol 2011). Using element channel data,virtual beam focusing can be used to reconstruct a large number of beamsfrom only a few acoustic interrogations. Furthermore, the rate of motion estimationcan be increased by using temporally unequispaced acquisition sequences(TUAS), a technique developed previously by our group, whichincreases the motion estimation rate while reducing the motion sampling rate.Methods—Using a Verasonics scanner with a custom TUASand virtual beam focusing, we have acquired short-axis views of human(n = 1) and open-chest canine (n = 3) hearts. Virtual beam focusing wasused to reconstruct 156 radiofrequency (RF) lines from 12 focusedacoustic transmits. Lateral displacement estimation was performed usingRF cross-correlation, and strain was computed using a least squares strainestimator. The quality of strain estimation was evaluated by using the elastographicsignal-to-noise ratio (SNRe).Results—We first show that lateral strain estimation can beachieved in vivo using virtual beam focusing and that the results are similarto conventional beam focusing. Also, we show that it is possible touse TUAS to estimate strain using a wide range of frame rates (57, 130,447, 894, 1788, and 2682 Hz) while maintaining this high beam density.The SNRe for lateral strain was found to range from 2 to 10 and be optimalat a motion estimation rate of 894 Hz.Conclusions—In conventional scanners, there is a trade-off betweenthe beam density and motion estimation rate, leading to a theoreticallimit on the quality of strain estimation. The use of virtual beamfocusing and TUAS techniques for lateral strain imaging was shown toeliminate this tradeoff and lead to good-quality lateral strain estimation athigh frame rates.