Image Enhancement by Noncontrast Harmonic Echocardiography. Part I. Qualitative Assessment of Endocardial Visualization

Original Article

Image Enhancement by Noncontrast Harmonic Echocardiography. Part I. Qualitative Assessment of Endocardial Visualization SHARON

L.

MULVAGH, M.D., DAVID

A. FOLEY, M.D.,

MAREK BELOHLAVEK, M.D., PH.D.,

AND JAMES B. SEWARD, M.D.

0.36) than for either fundamental mode (0.49 ± 0.21 and 0.57 ± 0.27 for the 2.0- and 3.5-MHz images, respectively). The harmonic images were ranked as better (P
• Objective: To determine whether harmonic imaging-use of signals with frequencies twice that of the transmitted ultrasound to produce ultrasound images--ean improve endocardial border definition in patients who have technically difficult echocardiograms. • Methods: We studied 29 patients with technically difficult echocardiograms (nonvisualization of 2 or more endocardial segments in a 16-segment model). Apical long-axis, four-chamber, and two-chamber images were acquired during fundamental imaging (at 2.0 and 3.5 MHz) and second harmonic imaging (3.5-MHz receive mode) in random order. Images were digitally stored and subsequently reviewed blindly for endocardial segment score (0 not visualized; 1 adequate; or 2 excellent) and overall ranking of image quality (1 [best] to 3 [worst]). • Results: Mean endocardial segment score was significantly better (P
=

=

=

ESS =endocardial segment score; F2 =conventional (fundamental) imaging at 2.0 MHz; F3.5 =conventional (fundamental) imaging at 3.5 MHz; H =second harmonic imaging

R

ecently, harmonic imaging has been introduced to augment detection of microbubble contrast agents. This technique uses broadband transducers and unique signal processing algorithms to detect signals with a frequency twice that which was emitted by the system (that is, "second harmonic" frequency). These harmonic frequencies are strongly produced by ultrasound contrast microbubbles.!" Backscattered ultrasound from insonated native tissue contains very weak harmonic signals in comparison with the signals at the emitted fundamental frequency, but the harmonic signals can be detected and used to generate images after sufficient amplification.'> Preliminary observations from clinical trials of microbubble contrast agents suggested improvement in the definition of cardiac structures with use of second harmonic imaging before administration of the contrast agent. These observations are consistent with in vitro measurements that have shown improved resolution with harmonic imaging.v' Therefore, we systematically evaluated the utility of second harmonic imaging for improving suboptimal echo-

cardiographic images. Specifically, we sought to demonstrate that endocardial border definition is substantially improved in patients undergoing rest echocardiography. PATIENTS AND METHODS

Study Population Twenty-nine adult patients referred for rest echocardiography and found to have technically difficult echocardiograms were studied. A technically difficult study was defined as absence of visualization of at least two endocardial segments in the standard 16-segment model" from the apical views. The 17 male and 12 female patients enrolled in the study had a mean weight of 85 kg (range, 54 to 144) and a mean body surface area of 2.0 m2 (range, 1.5 to 2.7). The Mayo Institutional Review Board had approved the study protocol, and informed consent was obtained from each patient.

Image Acquisition At completion of the routine, clinically indicated study and identification of suitably poor image quality, the investigational imaging was performed by using a Sequoia C256 (Acuson Corporation, Mountain View, California) with Native Tissue Harmonic software as follows. With the patient in the left lateral decubitus position, images were acquired from the apical four-chamber, long-axis, and two-

From the Division of Cardiovascular Diseases and Internal Medicine (S.L.M., DAF" M,B" J.B.S.) and Department of Physiology and Biophysics (M.B.), Mayo Clinic Rochester, Rochester, Minnesota. Address reprint requests to Dr. S. L. Mulvagh, Division of Cardiovascular Diseases, Mayo Clinic Rochester, 200 First Street SW, Rochester, MN 55905.

Mayo Clin Proc 1998;73:1062-1065

1062

© 1998 Mayo Foundation/or Medical Education and Research

For personal use, Mass reproduce only with permission from Mayo Clinic Proceedings,

Mayo Clin Proc, November 1998, Vol 73

chamber views. Three imaging modes were used: conventional (fundamental) imaging at two frequencies ("F2" [2.0 MHz] and "F3.5" [3.5 MHz]) and second harmonic imaging ("H" [1.75 MHz for transmitting and 3.5 MHz for receiving]). Acoustic power was set to maximum. Dynamic range was set at 70 dB. Edge enhancement and persistence were set to O. The preprocessing image contrast differentiation function, referred to as "delta" by the manufacturer, was set to 4 on the basis of preliminary visual optimization. The postprocessing contrast differentiation was set to 4 on the basis of selection of the least nonlinear curve that provided optimal image appearance. The "space-time" control was set to S\ to achieve balanced frame rate and temporal resolution. These imaging variables are similar to the standard preset values we use in our clinical echocardiography practice, based on our experience and advice from the manufacturer of the ultrasound system. In addition, we performed preliminary imaging on non-study subjects to ascertain that these settings provided acceptable images in all modes. The overall gain and time-gain compensation were optimized for each patient study and were not altered during acquisition. The transmit focus was consistently placed at the deepest farfield setting, to avoid any possible independent regional effect that its site may have on harmonic intensity formation across the ventricular myocardium. The cine-loop memory was used to capture the images in digital format, and they were subsequently stored to optical disk by using the imaging system's on-line "clip-store" software for later review. Image Analysis Digitally stored images were retrieved and displayed in full-screen format for assessment of endocardial definition. The cine-loop images obtained with the three imaging modes in each of the views were presented for each patient in random order to two experienced echocardiographic reviewers (S.L.M. and J.B.S.). The reviewers were "blinded" to the imaging mode and independently evaluated the images. Each of 18 segments (6 per view in 3 views, based on the American Society of Echocardiography 16-segmental models-s-that is, visualization of 2 apical segments in 2 views resulted in 18 total segments) was assigned an endocardial segment score (ESS) of 0 (not visualized), 1 (adequate), or 2 (excellent) based on the delineation of the endocardial surface throughout the cardiac cycle. A mean ESS encompassing the 3 views was calculated for each imaging mode by region (base, midventricle, and apex) and globally (all 18 segments). For each view, the cine-loop images were then displayed with all three imaging modes together in a quadscreen format. The position of the images was randomly

Assessment of Noncontrast Harmonic Imaging

1063

assigned. The observers were asked to rank the images on the basis of their overall quality from 1 (best) to 3 (worst). Statistical Analysis The ESSs were compared by repeated-measures analysis of variance. Differences in ESS between groups were evaluated by paired t tests. Rank scores were compared by using Friedman's test. Agreement of the two observers for segmental ESS and rank scores was evaluated by using the kappa statistic. The global mean ESSs for the two observers were also compared by Pearson correlation. P values of less than 0.05 were considered statistically significant. RESULTS Harmonic imaging subjectively improved endocardial visualization (Fig. 1). Mean ESS varied significantly across imaging mode (F = 171; P
For personal use. Mass reproduce only with permission from Mayo Clinic Proceedings.

1064 Assessment of Noncontrast Harmonic Imaging

Mayo Clin Proc, November 1998, Vol 73

Fig. 1. Representative appearance of fundamental imaging at 2.0 MHz (left), fundamental imaging at 3.5 MHz(center), and harmonic imaging (right) for the apical four-chamber (top) and two-chamber (bottom) echocardiographic views. of ultrasound equipment now include harmonic imaging as a standard feature on their newest imaging systems, primarily because of the demand for this technology for use with ultrasound contrast agents . Access to the second harmonic imaging package is generally simple , and the technique of harmonic imaging does not differ substantially from conventional imaging. System power outputs are the same in fundamental and harmonic imaging. Thus, the improved image quality is achieved without appreciable added effort or risk to the patient. Similar improvement in clinical images has been reported preliminarily by others.':" Our data are consistent with theoretical and in vitro work with harmonic imaging.P Ultrasound undergoes nonlinear distortion as it propagates through tissue; the result is the production of sound waves with frequencies at multiples of the applied ultrasound frequency (that is, harmonic frequencies). The magnitude of this distortion, and therefore the strength of the harmonic signals, is dependent on the amplitude of the applied ultrasound. The generation of harmonic signals increases with the distance traveled

through the tissue, except as limited by attenuation and dissipation of the fundamental ultrasound field, but the intensity remains well below that of the insonating beam. The harmonic signals thus generated form a beam coaxial to the fundamental beam. The second harmonic beam has significantly lower side lobes (relative to the strength of the main lobe) than the fundamental beam. The main lobe may also be narrower, depending on the imaging conditions. These features should improve the lateral component of contrast resolution and reduce imaging artifacts . Axial resolution is related to imaging frequency. Although imaging conventionally with a higher frequency should improve axial resolution, the increased attenuation associated with the higher frequency often precludes this approach in clinical practice . In our study, we found the harmonic images to be superior to the same-frequency fundamental (3.5-MHz) images. In the model described by Christopher,' the second harmonic beam had substantially lower side lobes than the same-frequency fundamental beam. Thus, the improved lateral contrast resolution seems more likely to be responsible for the observed image en-

For personal use. Mass reproduce only with permission from Mayo Clinic Proceedings.

Mayo Clin Proc, November 1998, Vol 73

Assessment of Noncontrast Harmonic Imaging

(see pages 1066 to 1070) explores one aspect of the mechanism for improved image quality with harmonic imaging.

Table I.-Endocardial Visualization Scores, Stratified by Imaging Mode and Region* Region Mode H

F2 F3.5

1065

Apex

Midventricle

Base

Global

0.73 0.27 0.44

1.16 0.59 0.63

1.15 0.61 0.66

1.02 0.49 0.57

*F2 and F3.5 =conventional (fundamental) imaging at 2 and 3.5 MHz, respectively; H = second harmonic imaging.

hancement by harmonic imaging than any effect of improved axial resolution from the conversion to a higher ultrasound frequency. Because of the increased generation of harmonic signals with propagation distance, we looked for regional variation in the second harmonic effect from apex to base. Although the image quality at the base and midventricle was better than at the apex for the harmonic images, this finding also prevailed for the fundamental images. Thus, we found no relationship between depth and improvement of image quality with second harmonic imaging, although we were unable to assess directly the effect of propagation distance on the second harmonic signals.

Study Limitations Our choice of machine settings was arbitrary, although guided by our clinical imaging experience and the recommendations of the system manufacturer. In fact, it would be impossible to compare the relative merits of each of the 1,200 combinations of delta, postprocessing curves, edge enhancement, and space-time resolution offered by the equipment we used, let alone evaluate the additional effects of dynamic range, output power, and transmit focus. Although the machine settings inadvertently may have favored one imaging mode over another, we thought that acceptably good images were observed in all modes in our preliminary tests, and the settings approximate those used daily in our clinical practice. In this small group of patients, we were unable to evaluate the clinical significance of the improved endocardial definition (that is, new diagnostic information). We also did not address whether the improved resolution is applicable to evaluation of other cardiac structures. Although the reviewers were "blinded" to the imaging mode, the differences in visual appearance of the three imaging modes make this blinding questionable-a challenge shared by subjective comparisons of all imaging techniques. Although the current data are concordant with prior theoretical and laboratory work, this study did not attempt to examine further the mechanisms by which harmonic imaging improves the quality of images. A companion article

CONCLUSION We have demonstrated that noncontrast harmonic imaging substantially enhances endocardial definition in patients with suboptimal conventional echocardiographic images and considerably improves overall subjective image quality. Harmonic imaging can be done without added patient risk or study time. The hope is that the improvements in endocardial definition seen with harmonic imaging will result in enhanced diagnostic accuracy and extension of echocardiographic techniques to patients with marginal acoustic windows. ACKNOWLEDGMENT We thank Gregory Gilman, R.N., R.D.C.S., Geralyn M. Pumper, R.N., R.D.C.S., and Joan L. Jensen, R.D.C.S., who were instrumental in the echocardiographic image acquisition.

REFERENCES 1. 2. 3.

4. 5. 6.

7.

8.

9. 10. 11. 12.

Miller DL. Ultrasonic detection of resonant cavitation bubbles in a flow tube by their second-harmonic emissions. Ultrasonics 1981;19:217224 de Jong N, Ten Cate FJ, Lancee CT, Roelandt JR, Bom N. Principles and recent developments in ultrasound contrast agents. Ultrasonics 1991;29:324-330 Mulvagh SL, Foley DA, Aeschbacher BC, Klarich KK, Seward JB. Second harmonic imaging of intravenously administered echocardiographic contrast agent: visualization of coronary arteries and measurement of coronary blood flow. J Am Coli Cardiol 1996;27:1519-1525 Christopher T. Finite amplitude distortion-based inhomogeneous pulse echo ultrasonic imaging. IEEE Trans Ultrason Ferroetectncs FrequencyControl 1997;44:125-139 Ward B, Baker AC, Humphrey VF. Nonlinear propagation applied to the improvement of resolution in diagnostic medical ultrasound. J Acoust Soc Am 1997;101:143-154 Schiller NB, Shah PM, Crawford M, DeMaria A, Devereux R, Feigenbaum H, et al (American Society of Echocardlography Committee on Standards, Subcommittee on Quantltatlon of Two-Dimensional Echocardiograms). Recommendations for quantification of the left ventricle by two-dimensional echocardiography. J Am Soc Echocsrdlogr 1989;2:358-367 Kasprzak JD, Paelinck B, Vletter WB, de Jong N, Ten Cate FJ. Second harmonic imaging of left ventricle improves detection of endocardium with and without contrast agent [abstract]. Circulation 1997;96(Suppl 1):1-148 Bednarz J, Spencer KT, Mor-Avi V, Korcarz CE, Rafter P, Lang RM. Can harmonic imaging without contrast enhancement aid in the echocardiographic evaluation of left ventricular function? [abstract]. Circulation 1997;96(Suppl 1):1-584 Tiemann K, Schlosser T, Likungu J, Koester J, Blmmel D, Becher H. Harmonic imaging without contrast-does blood or tissue emit harmonic frequencies? [abstract]. Circulation 1997;96(SuppI1):1-584 Senior R, Hayes T, Powers J, Khattar RS, Lahlrl A. Improved endocardial visualisation with second harmonic compared to conventional 2D imaging [abstract]. Circulation 1997;96(Suppl 1):1-584 Pyles JM, Sawada SG, Feigenbaum H, Segar DS. Enhancedendocardial visualization using harmonic imaging without contrast [abstract]. Circulation 1997;96(Suppl 1):1-585 Rubin DN, Yazbek N, Homa D, Odabashlan J, Livisi B, Stewart WJ, et al. Why does tissue harmonic imaging improve image quality: a quantitative examination demonstrating side-lobe suppression [abstract]. J Am Coli Cardiol 1998;31(Suppl A):127A

For personal use. Mass reproduce only with permission from Mayo Clinic Proceedings.