# Quantitative assessment of damage during MCET: a parametric study in a rodent model

- Yiying I. Zhu
^{1, 2}, - Douglas L. Miller
^{2}, - Chunyan Dou
^{2}, - Xiaofang Lu
^{1}and - Oliver D. Kripfgans
^{1, 2}Email author

**3**:18

**DOI: **10.1186/s40349-015-0039-2

© Zhu et al. 2015

**Received: **12 March 2015

**Accepted: **7 October 2015

**Published: **16 October 2015

## Abstract

### Background

Myocardial cavitation-enabled therapy (MCET) has been proposed as a means to achieve minimally invasive myocardial reduction using ultrasound to produce scattered microlesions by cavitating contrast agent microbubbles.

### Methods

Rats were treated using burst mode focused ultrasound at 1.5 MHz center frequency and varying envelope and pressure amplitudes. Evans blue staining indicated lethal cardiomyocytic injury. A previously developed quantitative scheme, evaluating the histologic treatment results, provides an insightful analysis for MCET treatment parameters. Such include ultrasound exposure amplitude and pulse modulation, contrast agent dose, and infusion rate.

### Results

The quantitative method overcomes the limitation of visual scoring and works for a large dynamic range of treatment impact. Macrolesions are generated as an accumulation of probability driven microlesion formations. Macrolesions grow radially with radii from 0.1 to 1.6 mm as the ultrasound exposure amplitude (peak negative) increases from 2 to 4 MPa. To shorten treatment time, a swept beam was investigated and found to generate an acceptable macrolesion volume of about 40 μL for a single beam position.

### Conclusions

Ultrasound parameters and administration of microbubbles directly influence lesion characteristics such as microlesion density and macrolesion dimension. For lesion generation planning, control of MCET is crucial, especially when targeting larger pre-clinical models.

### Keywords

Cavitation microlesions Hypertrophic cardiomyopathy Myocardial macrolesion Therapeutic ultrasound Quantitative therapy analysis## Introduction

Hypertrophic cardiomyopathy (HCM) is a common genetic cardiovascular disease, which is usually clinically recognized by a maximal left ventricular wall thickness greater than 15 mm [1]. This globally prevalent disease, reported in about 0.2 % (i.e., 1:500) of the general population, is the most frequent cause of sudden death in young people and can lead to functional disability from heart failure and stroke [2].

The traditional treatment for HCM to reduce myocardium is septal myectomy. This surgical method removes septal hypertrophy, which possibly leads to perturbation of mitral valve leaflets [3]. An innovative therapeutic scheme, named myocardial cavitation-enabled therapy (MCET), has been proposed as a means to achieve minimally invasive myocardial reduction by cavitating contrast agent microbubbles with ultrasound to produce a fractional macrolesion containing sparse and histologically definable microlesions [4]. There are several ways of controlling cavitation here. Cavitation is enabled by the injection of ultrasound contrast agents. These will enable cavitation only in the focal region of the transducer and thus only there lead to microlesion formation in the myocardium. Second, ultrasound cavitation is dependent on sound pressure amplitude. In vivo experiments reveal that cavitation-induced lesions take place at peak rarefactional pressures larger than 2 MPa as obtained under free field conditions. In this case, ECG is monitored for premature complexes. It has been seen that the occurrence of premature complexes is directly correlated with cavitation events [5].

As a potential tissue reduction therapy, MCET avoids open-chest surgery and is hypothesized to allow healing with minimal scar formation, resulting in shrinkage of the cardiac treatment volume. This ultrasound microbubble-enabled method additionally provides the possibility of guiding and monitoring via quantifying feedback from the microbubble emissions.

To optimize MCET ultrasound parameters and administration of microbubble settings, assessment of the therapeutic effect is needed to assist parameter adjustment. Efforts in computerized analysis have been made to aid diagnostics and therapy for being fast, objective, and quantitative. Methods have been developed for computed tomographic angiography for the purposes of detecting heart diseases [6, 7] and for quantification of coronary arterial stenosis [8]. Automatic detection of pulmonary embolism has also been used in CT angiography [9, 10]. Quantitative ultrasound has been employed in diagnosis of osteoporosis [11], as well as in at-risk pregnancies with three-dimensional sonographic measurement of blood volume flow in umbilical cords [12]. Three-dimensional high-frequency ultrasound data also has been processed to offer a quantitative evaluation of cancerous lymph nodes at the microscopic level [13].

For MCET, a quantitative method for assessing the distribution and total accumulation of myocardial necrosis based on Evans blue-stained cells in the tissue histology slices was developed previously [14] and is used in this study. This paper investigates the tuning of various parameters involved in MCET and paves the way for pre-clinical treatment planning of myocardial lesion creation and properties thereof, in a quantitative manner.

One important and practical aspect of MCET is managing the buildup of microlesions and macrolesions to achieve a desired amount of myocardium reduction in larger pre-clinical models as well as, ultimately, in the clinic. Acoustic pressure amplitude, contrast dose, and treatment duration are adjustable variables. The parametric exploration of various conditions will assist in the search for feasible treatment conditions that allow for fast lesion creation with a 15–20 % microlesion density and a large axial and lateral dimension. Another desirable factor for practical clinical implementation is the treatment efficiency. Instead of treating a single focal spot as done in our previous study [4], a scanned beam would allow for a more rapid accumulation of lesions in a larger target treatment volume.

Our method of computer-aided histology analysis was developed using relatively high exposure parameters to reflect therapeutic treatment conditions [12]. This provided a means to reconstruct the tissue volume containing microlesions and their distribution, which can then be integrated to yield the potential fraction of tissue reduction. For validation, a visual scoring method was used in tandem, in which lethally injured cells indicated by fluorescent staining in frozen sections were counted. The visual method has been the gold standard for quantifying cell death by counting the absolute number of stained cells. However, when the number of stained cells becomes large, as for treatment (rather than exploring bioeffects), the visual method becomes a qualitative scoring method, which was suspected to yield inaccurate results for the validation for the computer-aided method. The purpose of this study was to analyze several exposure groups, which had reduced, sub-therapeutic treatment effects, using quantitative visual scoring for comparison to the computer-aided analysis.

## Materials and methods

### Experimental conditions

Table of sets of conditions used for respective groups of rats, with a cohort of five animals each

Experiment conditions for rat groups | |||||
---|---|---|---|---|---|

Group ID | Infusion site | Pressure (PRPA) | Modulation | Infusion rate | Treatment duration |

A | Tail | 4 MPa | Square | 5 μL/kg/min | 5 min |

B | Jugular | 4 MPa | Square | 5 μL/kg/min | 5 min |

C | Jugular | 4 MPa | Square | 5 μL/kg/min | 5 min |

D | Jugular | 4 MPa | Gaussian | 5 μL/kg/min | 5 min |

E | Jugular | 4 MPa | Gaussian | 12.5 μL/kg/min | 2 min |

G | Tail | 2 MPa | Gaussian | 5 μL/kg/min | 5 min |

H | Tail | 2.8 MPa | Gaussian | 5 μL/kg/min | 5 min |

I | Tail | 4 MPa | Gaussian | 5 μL/kg/min | 100 s |

J | Tail | 4 MPa | Gaussian | 5 μL/kg/min | 30 s |

Different contrast agent dose rates were tested by comparing groups D and E. The previous rate was 5 μL/kg/min, representing the recommended dose for diagnostic applications [4]. A higher infusion rate of 12.5 μL/kg/min was tested for the possibility of using a higher dose in therapeutic applications, which may reduce treatment durations.

Comparison between groups G, H, and D evaluated the dependence of lesion formation on acoustic pressure. The three groups were respectively exposed to ultrasound fields of 2, 2.8, and 4 MPa PRPA. Groups J, I, and D, on the other hand, evaluated microlesion accumulation by varying the treatment duration, i.e., adjusting the contrast infusion duration. Groups G, H, I, and J were specifically treated with sub-therapeutic parameters with reduced treatment impact on cell survival. Correlation between acoustic pressure, treatment duration, and induced microlesion density was intended to establish some dynamic range for microlesion induction.

Finally, group F was a sham and calibration control group, in which each rat received the full 4 MPa therapy exposure before the contrast agent infusion started.

Results in groups are presented in boxplots. For each box, the central mark is the median, the edges of the box are the 25th and 75th percentiles, the whiskers extend to the most extreme data points not considered outliers, and outliers are plotted individually. The normal range was defined as *q*
_{3} + 1.5 (*q*
_{3} − *q*
_{1}) or smaller than *q*
_{1} − 1.5 (*q*
_{3} − *q*
_{1}), where *q*
_{1} and *q*
_{3} are the 25th and 75th percentiles, respectively.

### Cardiomyocyte scoring

Rat hearts were harvested and scored 1 day after exposure as described in previous work [15]. Briefly, up to 40 10-μm-thick frozen sections were made from the treated volume in each heart. A quantitative method for assessing the distribution and total accumulation of myocardial necrosis is based on Evans blue staining and was developed previously [14]. Microlesions were identified by fluorescence microscopy and photographs of each section. Image registration was then performed to digitally stack the frozen sections in 3D and to reconstruct a model of the heart morphology in the entire sampled region showing the three-dimensional distribution of microlesions. The microlesion fraction of the tissue within the focal zone was calculated to estimate the potential fractional volume of tissue reduction that was achieved. Quantitative results were characterized in terms of microlesion volume, macrolesion volume, microlesion lesion density, and dimensions of the radially symmetric approximated macrolesion.

In addition to the computer-aided assessment, traditional visual scoring was used to evaluate myocardial necrosis qualitatively by visual identification and quantitatively by scoring of Evans blue-stained cells using fluorescence microscopy [15]. Automatic scores were obtained from dividing the geometric microlesion volume by a constant conversion factor acquired from a geometry-based cardiomyocyte model [14].

### Therapeutic field simulation

*x*= 0 mm) axis, the effective volume, denoted as an axisymmetric rotational model, was calculated. This simulation intended to associate in situ acoustic field with the formed lesion.

## Results

### Quantitative computer-guided lesion analysis

*y*=

*a*

_{1}·

*x*, was performed for the low lesion count segment and is shown in red and extended as blue in Fig. 4. Another linear least square fit, now of the form

*y*=

*a*

_{2}·

*x*+

*b*, was performed for the high lesion count segment and is shown in red in Fig. 4. Best-fit coefficients were found as:

*a*

_{1}= 1.52,

*a*

_{2}=

*0.19*and

*b*= 15,216, with

*a*

_{1}greater than 1 would mean that visual scoring counted more cells,

*a*

_{2}being so low shows that visual scoring counted groups of cells as one. This result clearly showed the “saturation” effect of the visual scoring method for high cell counts.

### Acoustic pressure dependence

#### Macrolesion volumes

*p*

_{L}required for lesion formation, will contribute to the therapy. This volumetric region grows as the pressure amplitude at the focus grows, and thus, larger acoustic pressures yield larger lesion count. Macrolesion dimensions are also dictated by the spatial availability of cells and contrast agent. The simulation gave an acoustic field of 2.5 × 32 mm at the level of −6 dB relative to 4 MPa as shown in Fig. 3b, representing the expected lesion region. In vivo results for the parametric acoustic pressure amplitude cases 2.0 (G), 2.8 (H), and 4.0 MPa (D) are illustrated in Fig. 6 using boxplots. Frequently, the axial dimension was limited by the thickness of the myocardium with the therapy beam penetrating the left ventricle entirely before the pressure amplitude fell below the threshold

*p*

_{L}. Likewise, the therapy beam pressure rose above

*p*

_{L}before entering the myocardium. Therefore, an increase in pressure at the focus will only show limited increase in the axial lesion size. Radial macrolesion expansion is hyperlinear and follows the prediction of the simulation shown as the plotted curve in Fig. 6c. Simulations were done to compute the average radius of the acoustic beam above the hypothesized threshold

*p*

_{L}(2.0 MPa).

#### Microlesion characteristics

*y*= 4.52

*x*− 4.07, with 0 % lesion density occurring at 0.9 MPa indicated by the red circle. The microlesion volumes for resliced disks versus different acoustic amplitudes are plotted in Fig. 7b with least square fitting resulting in

*y*= 57.2

*x*− 130.0 and zero microlesion volume occurring at 2.3 MPa indicated by the red circle.

### Contrast agent availability versus macrolesion characteristics

As alluded to in the previous section, macrolesion dimensions are dictated by several experimental conditions, including contrast agent availability. More available agent will likely generate more lesions. On the other hand, more contrast agent per unit time may lead to agent-induced acoustic shadowing and a diminished in situ pressure wave amplitude. Results for changes in contrast agent availability are shown next.

#### Infusion duration

#### Infusion rate

*microlesion volume*here refers to the total volume of all induced microlesions over the entire myocardium, which is not identical to the shown microlesion volume within the characterized macrolesion (macrolesion volume times lesion density).

### Therapy beam sweeping

The pulse modulation groups corresponding to square and Gaussian profile are shown in Fig. 2. In the simulation modeling, the volume exposed by negative pressures greater than 2 MPa, and marked by circles, were integrated across space and time yielding 4.5 and 2.7 μL·s for square and Gaussian modulations, respectively.

## Discussion

The quantitative results generated by the previously developed and here tested computer-aided scheme provide possibilities for numeric and quantitative 3D lesion analysis and their dependence on experimental parameters that were investigated for their relevance for developing and improving MCET.

### Cardiomyocyte scoring

### Experimental perturbations exclusions

*does not*impact the therapeutic result. In vivo variations can be significant (see error bars in Fig. 12); however, there is a significant overlap between the tested routes for injection and between various aiming paths.

### Acoustic field modeling

#### Acoustic amplitude

Acoustic modeling provides a way to approximate lesion formation. In the therapy, impact analysis bioeffects of various acoustic exposures were investigated. The associated field simulation underestimates the mean radius of the macrolesion at 4 MPa as shown in Fig. 6c. This is because in vivo, some of the beam penetrates the left ventricle. This part of the beam was not excluded in the simulation; thus, the simulation resulted in averaging out the affected macrolesion radius. Another factor contributing to the greater treatment effect seen in vivo than the simulation comes from deformation of hearts after being harvested. Rats were treated at the ends of systole but hearts were relaxed after being sacrificed. Thus, the acoustic pattern may be distorted to some extent.

#### Swept beam

Treatment in humans will require focusing of the therapeutic beam at a larger area than currently done in rodents (rats). Such will either lead to the need of a modified, i.e., larger, point spread function or more likely numerous repetitions of individual exposures. The latter can be realized by either individual focal treatments or by employing a swept beam. The former will require a longer time for treatment of an equivalent total count of focal spots since the beam will be stepped from treatment location *n* to *n* + 1; therefore, a swept beam was investigated. Illustrated by “O” plot marks in Fig. 2 are the individual tone bursts that exceed the lesion formation pressure threshold. With that and the previously mentioned axisymmetric rotational model (point spread function), an effective treatment volume was simulated. Integration across space and time, yielded 4.5 and 2.7 μL·s for square and Gaussian modulations, respectively. Therefore, the swept beam simulation predicts an effective volume of 59.8 % of that of an individual focal treatment. The experimental macrolesion volume shown in Fig. 10a yields an effective median volume fraction of 61.5 %, supporting the axisymmetric rotational volumetric model.

### Thresholded induced and statistically accumulated lesion

#### Lesion formation as accumulated statistical events

In the experiment of increasing infusion, the slightly decreasing trend of lesion density shown in Fig. 6 may indicate some shadowing effect caused by a large population of instantaneous microbubbles placed along the beam path, acting as scatterers. However, the shadowing factor will need to be verified for the cases of a longer path, such as in a larger animal model.

#### Microlesion density versus ultrasound amplitude

Least square fitting in Fig. 7a implies that an increasing acoustic pressure has a positive effect on microlesion formation up to 4 MPa, i.e., higher acoustic pressure or exposure possesses the higher potential to induce bioeffects. A similar positive correlation of acoustic amplitude on cavitation-induced bioeffects was also presented by Samuel et al. [18]. One possible reason is that larger pressure will result in a larger active microbubble population. Therefore, a larger density of microlesion will be induced by the then more frequent microbubble cavitation events. For a constant pressure amplitude, macrolesions also grow radially over time. This is because of the probabilistic accumulation of microlesions on the penumbra of the current macrolesion. There, the probability *χ* for microlesion generation is larger than 0 % and smaller than 100 %, i.e., the sound pressure amplitude is close to the threshold *p*
_{L} discussed above. If *χ* is 20 %, then a five times longer exposure will statistically result in additional lesion formation.

#### Acoustic pressure threshold

The zero microlesion volume shown in the line fit in Fig. 7b might overestimate the acoustic pressure threshold for lesion induction. This is because microlesion volumes were characterized in a way that could bias towards beam regions with lower acoustic pressure and hence lower partial microlesion volume.

Additionally, the acoustic pressure threshold for microlesion induction by either lesion densities or microlesion volumes is a rough estimation and may be inaccurate due to biological variations, with *R*
^{2} being 0.48 and 0.56.

### Application in human

Ideally, the wanted axial length of the transducer’s point spread function matches the myocardial thickness. Here in the presented small animal model, a shorter depth of field would have been desirable, though no side effects, except for some pulmonary hemorrhaging, presented in the study. The characterized macrolesions for all rats showed similar lengths, which in most cases here is due to the acoustic path restrained by the limited thickness of the left vertical wall. However, when the study moves to a larger animal model (such as swine) rather than the currently employed rodents (rats), the effect of therapeutic pressure on the macrolesion length will manifest itself and be vital for lesion formation and accretion.

The shown results serve as a preliminary test of MCET for application in humans. The main benefit from this therapy method is the minimal invasiveness and the hypothesis that cavitation-induced, sparsely distributed microlesions do not lead to major infarct-like scars, which can disrupt conduction pathways and lead to heart block, such as for alcohol ablation treatment [19]. The relationship between lesion characteristics from the small animal model, such as lesion dimension and density and in situ ultrasound field, is assumed to be analogous to that of a large animal model. Both models follow the same rationale of lesion formation. In the larger model, especially in humans, it is anticipated that we will have to create composite lesions, i.e., lesions created by electronically and/or mechanically sweeping the beam. For therapy of large volumes, the therapy beam will likely be scanned through the desired volume (simulated here by the Gaussian modulation) to accomplish the treatment in less time than is needed to treat point by point (as in HIFU). Suppose a human subject needs MCET treatment at a myocardial region of 4 cm diameter [20]. Assuming that the left ventricular wall of the hypertrophic heart has a 21-mm thickness [2], a macrolesion, approximated as a cylinder, with volume π × (4 cm/2)^{2} × 21 mm = 26.4 mL, would be needed. Assuming that the therapy employs the recommended dose of Definity® for diagnostic exams, i.e., 5 μL/kg/min, a 5-min treatment of a single focal spot will yield a 50-μL macrolesion with 20 % microlesion density. To create the aforementioned macrolesion, a total duration of approximately 2640 min will be needed to achieve the desired lesion volume. Recruitment of a swept beam to foster the lateral lesion formation at 56 fps, as discussed above, will reduce the total duration to 47 min. Stacking multiple axial focal zones will enlarge axial lesion size and further accelerate the therapy. The above calculations are under the assumption that the in vivo microbubble distribution in the myocardium is similar in human and the chosen rat model. A large animal model for human cardiophysiology, such as swine, will be needed for investigating the clinical transition for MCET.

## Conclusion

The quantitative scoring scheme overcomes the limitation of traditional visual scoring and works for histological cases with a large lesion count, i.e., has an appropriate dynamic range for evaluating therapeutic applications. The presented results have shown that MCET-induced macrolesions grow radially as the acoustic pressure amplitude increases. A swept beam as a new method to shorten treatment time seems promising but requires additional verification to ensure efficacy. These characterizations and validations may assist future MCET treatment planning.

## Declarations

### Acknowledgements

This work was supported by PHS grant HL114595 awarded by the National Institutes of Health, DHHS.

**Open Access**This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

## Authors’ Affiliations

## References

- Gersh BJ, Maron BJ, Bonow RO, Dearani JA, Fifer MA, Link MS, et al. 2011 ACCF/AHA guideline for the diagnosis and treatment of hypertrophic cardiomyopathy: executive summary: a report of the American college of cardiology foundation/American heart association task force on practice guidelines. J Am Coll Cardiol. 2011;58(25):2703–38.View ArticlePubMedGoogle Scholar
- Maron BJ, Maron MS. Hypertrophic cardiomyopathy. Lancet. 2013;381(9862):242–55.View ArticlePubMedGoogle Scholar
- Marian AJ. Contemporary treatment of hypertrophic cardiomyopathy. Tex Heart Inst J. 2009;36(3):194–204.PubMed CentralPubMedGoogle Scholar
- Miller DL, Dou C, Owens GE, Kripfgans OD. Optimization of ultrasound parameters of myocardial cavitation microlesions for therapeutic application. Ultrasound Med Biol. 2014;40(6):1228–36.PubMed CentralView ArticlePubMedGoogle Scholar
- Miller DL, Dou CY, Lucchesi BR. Are Ecg premature complexes induced by ultrasonic cavitation electrophysiological responses to irreversible cardiomyocyte injury? Ultrasound Med Biol. 2011;37(2):312–20.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhou C, Chan HP, Chughtai A, Patel S, Hadjiiski LM, Wei J, et al. Automated coronary artery tree extraction in coronary CT angiography using a multiscale enhancement and dynamic balloon tracking (MSCAR-DBT) method. Comput Med Imaging Graph. 2012;36(1):1–10.PubMed CentralView ArticlePubMedGoogle Scholar
- Bouraoui B, Ronse C, Baruthio J, Passat N, Germain P. 3D segmentation of coronary arteries based on advanced mathematical morphology techniques. Comput Med Imaging Graph. 2010;34(5):377–87.View ArticlePubMedGoogle Scholar
- Xu Y, Liang GY, Hu GS, Yang Y, Geng JZ, Saha PK. Quantification of coronary arterial stenoses in CTA using fuzzy distance transform. Comput Med Imaging Graph. 2012;36(1):11–24.View ArticlePubMedGoogle Scholar
- Zhou C, Hadjiiski LM, Sahiner B, Chan HP, Patel S, Cascade PN, et al. Computerized detection of pulmonary embolism in 3D computed tomographic (CT) images: vessel tracking and segmentation techniques. P Soc Photo Opt Ins. 2003;5032:1613–20.Google Scholar
- Bouma H, Sonnemans JJ, Vilanova A, Gerritsen FA. Automatic detection of pulmonary embolism in CTA images. IEEE Trans Med Imaging. 2009;28(8):1223–30.View ArticlePubMedGoogle Scholar
- Moayyeri A, Adams JE, Adler RA, Krieg MA, Hans D, Compston J, et al. Quantitative ultrasound of the heel and fracture risk assessment: an updated meta-analysis. Osteoporosis Int. 2012;23(1):143–53.View ArticleGoogle Scholar
- Pinter SZ, Rubin JM, Kripfgans OD, Treadwell MC, Romero VC, Richards MS, et al. Three-dimensional sonographic measurement of blood volume flow in the umbilical cord. J Ultrasound Med. 2012;31(12):1927–34.PubMed CentralPubMedGoogle Scholar
- Mamou J, Coron A, Hata M, Machi J, Yanagihara E, Laugier P, et al. Three-dimensional high-frequency characterization of cancerous lymph nodes. Ultrasound Med Biol. 2010;36(3):361–75.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhu YI, Miller DL, Dou C, Kripfgans OD. Characterization of macrolesions induced by myocardial cavitation-enabled therapy. IEEE Trans Biomed Eng. 2015;62(2):717–27.PubMed CentralView ArticlePubMedGoogle Scholar
- Miller DL, Dou Y, Lu F, Zhu YI, Fabiilli ML, Owens GE, et al. Use of theranostic strategies in myocardial cavitation-enabled therapy. Ultrasound Med Biol. 2015;41(7):1865–75.View ArticlePubMedGoogle Scholar
- Miller DL, Dou CY, Wiggins RC. Simulation of diagnostic ultrasound image pulse sequences in cavitation bioeffects research. J Acoust Soc Am. 2007;122(4):2002–8.View ArticlePubMedGoogle Scholar
- Jensen JA, Svendsen NB. Calculation of pressure fields from arbitrarily shaped, apodized, and excited ultrasound transducers. IEEE Trans Ultrason Ferroelectr Freq Control. 1992;39(2):262–7.View ArticlePubMedGoogle Scholar
- Samuel S, Cooper MA, Bull JL, Fowlkes JB, Miller DL. An ex vivo study of the correlation between acoustic emission and microvascular damage. Ultrasound Med Biol. 2009;35(9):1574–86.PubMed CentralView ArticlePubMedGoogle Scholar
- ten Cate FJ, Soliman OI, Michels M, Theuns DA, de Jong PL, Geleijnse ML, et al. Long-term outcome of alcohol septal ablation in patients with obstructive hypertrophic cardiomyopathy: a word of caution. Circ Heart Fail. 2010;3(3):362–9.View ArticlePubMedGoogle Scholar
- Maron BJ. Surgery for hypertrophic obstructive cardiomyopathy: alive and quite well. Circulation. 2005;111(16):2016–8.View ArticlePubMedGoogle Scholar