Two aspects of micro-bubbles are important – their gas content and the nature of their shell. Recently-approved micro-bubbles almost universally involve encapsulation of a high molecular weight gas, which improves the persistence of the bubble, optimising the number available in the left heart chambers. Air has a greater propensity to dissolve into solution and although currently unattractive because of loss of gas before arrival on the right side of the heart, better encapsulation may allow its resurgence – the benefit would be more rapid disappearance when the bubble bursts. The nature of the shell or surface modifying agent, which improves stability and prevents dissolution, may become important for new targeted imaging approaches.

Even when optimal microbubble delivery to the myocardium is achieved with invasive coronary 1 or aortic root 2 injections, detection of reliable myocardial opacification using standard 2D imaging is difficult. Whilst this partly reflects difficulty distinguishing bright, grey scale echo signals from the myocardial tissue from those of micro-bubbles within the myocardial micro-circulation, the problem is multi-factorial and has been overcome by the development of contrast specific imaging modalities, which exploit the unique interaction between the ultrasound field and micro bubbles3 to maximise the received contrast backscatter and minimise myocardial tissue backscatter.

Micro-bubbles oscillate (expand and contract) in the ultrasound field. The pattern and nature of their oscillation, and thus the nature of the backscatter signal, differs, depending on the acoustic power of the transmitted ultrasound field, which is expressed on modern ultrasound machines as the mechanical index (MI). In general echocardiography, the amplitude of returning backscatter depends on the nature of the insonated structure, and is represented by brightness on the formed image. Most backscatter returns at the same frequency as the transmitted ultrasound (fundamental frequency).

With very low mechanical index imaging (MI < 0.1), micro bubbles demonstrate linear oscillation, where the contraction and expansion of the micro bubble are equal. All returning micro bubble backscatter remains within the range of the frequency transmitted by the transducer (the fundamental frequency) – as for surrounding structures (Figure 1a).

Low mechanical index imaging (0.1 – 0.3), generates nonlinear oscillation of the micro bubble whereby expansion is greater than contraction. In addition to the usual backscatter of a particular amplitude within the range of the fundamental frequency, the bubbles also produce backscatter of (lesser) amplitude at harmonic frequencies 45. For example scatter from a bubble interrogated by a 4 MHz pulse will be composed of 4 MHz and 8 MHz sound (Figure 1b). Tissues and surrounding structures do not produce harmonic backscatter when imaged under low power.

When exposed to high mechanical index imaging (MI > 0.6, ie the MI used for standard imaging) the bubbles oscillate wildly and burst. Upon destruction, micro-bubbles produce a brief, high amplitude signal, with good resolution, which is rich in harmonic signals, containing backscatter at the second third and fourth harmonics etc (Figure 1c). The bubble dissipates gradually following destruction, depending upon the gas composition of the bubble, with air-based micro-bubbles dissipating much more rapidly. Importantly at high MI, tissues also produce harmonics.

With standard fundamental imaging (high MI), myocardial micro-bubbles are destroyed continuously, causing apical swirling on LVO images and never replenishing the myocardium within the beam, and thus not returning significant myocardial backscatter. This exacerbates the underlying difficulty of differentiating on grey scale, between tissue backscatter and micro-bubble backscatter.

Changing to standard harmonic imaging (processing only signals returned at the second harmonic of the fundamental frequency) significantly improves LVO, but offers little benefit for myocardial perfusion, with ongoing bubble destruction and sub-optimal differentiation between contrast and tissue, because both produce harmonics at high power. Contrast specific imaging modalities, with the aim of enhancing the contrast to tissue backscatter signal ratio (CTR) are required for optimal performance of a myocardial contrast study.

The decision regarding the use of a bolus or infusion for intravenous administration of micro-bubbles is dependant on a variety of factors, including the type of micro-bubble used, equipment and staff available and clinical indication for the test (LVO, qualitative perfusion or quantitative perfusion). For MCE, bubbles should be infused with the aim of LV opacification and adequate myocardial perfusion with minimal/no attenuation of the basal segments. The amount of contrast required for this varies from bubble to bubble, machine to machine and technique used (intermittent or continuous), as well as showing patient to patient variability. Ideally all studies, particularly those assessing perfusion would involve a continuous infusion of micro-bubbles, with a mechanically controlled infusion preferable to a manually controlled one (slow continuous injection). This enables establishment of a true steady state for optimal imaging and in particular quantification.

While there is clear evidence that contrast LVO can improve endocardial border definition, there are theoretical reasons why this may not necessarily translate into more accurate assessment of LV cavity size, volume and function. Attenuation artefacts sometimes obscure the endocardial surface, particularly in the basal segments and myocardial contrast may confuse distinction of the border between the blood pool and the endocardium. Most importantly, the use of any 2-D echo technique to assess left ventricular volumes and ejection fraction (with or without contrast), requires a standard imaging plane. Fore-shortened views or views not oriented through the centre of the ventricular cavity will lead to an underestimation of volume no matter how good the endocardial border definition 16.

Notwithstanding these potential technical limitations, contrast LVO using fundamental imaging improves the correlation between LV volumes obtained with echo and MRI 17. In addition, accurate classification of systolic function by calculated ejection fraction has been improved from 70 to 94% by the use of contrast, with this improved accuracy almost exclusive to patients with poor baseline images. A subsequent study confirmed that, even using harmonic imaging, the measurement of LV volumes was optimised with contrast LVO, using electron beam CT as the standard of reference 18. While the role of LVO for 3D imaging is currently undefined, it seems likely that this will be an important application of contrast echo (Figure 6 and additional files 6, 7, 8 and 9).

Enhanced endocardial border definition can improve the accuracy and inter-observer agreement for assessment of regional wall motion at rest 19. Compared with MRI, the number of segments visualized was clearly improved after contrast (86% visualised before contrast, 99% after contrast). In addition, identification of segments with abnormal wall motion improved 82% to 100%, with the clinical utility of the contrast being greatest at the lateral and anterior walls. Importantly, the inter-observer agreement for assessment of individual wall segments was significantly improved and contrast also improved intra-observer agreement for determination of normal versus abnormal wall motion and assessment of the severity of wall motion abnormality compared with MRI. The ability of LVO to improve scoring and inter-observer variability of regional wall motion at rest has important implications for stress echocardiography.

Stress echocardiography is an established clinical tool with a high sensitivity and specificity for the diagnosis of coronary artery disease (CAD). During stress echo the diagnosis of CAD is based on detection of regional contractile dysfunction, and requires visualization of all myocardial segments to document or exclude abnormalities definitively. Reduced endocardial border definition is exacerbated during stress because of chest wall movement during hyperventilation and cardiac translational movement during tachycardia. With fundamental imaging, inadequate endocardial definition has been reported in up to 30% of stress echos 20. In addition, Hoffman et al demonstrated that suboptimal studies have worse reproducibility and a poorer inter observer variability, with inter-institutional institutional observer agreement as low as 43% for studies with poor image quality21 Tissue harmonic imaging, digital side by side analysis and standardised reporting criteria have alleviated but not overcome this problem22.

Wall motion scoring and reproducibility during stress echo were even improved with air filled contrast agents and fundamental imaging23. Perflurocarbon filled agents demonstrated almost complete and consistent endocardial border definition 24, with superiority even to tissue harmonic imaging, and the greatest improvement being seen in patients with poorest image quality (26).

Despite these clear advantages, the critical clinical question of whether LVO actually improves the accuracy of stress echo for diagnosis of CAD remains unanswered – Dolan 25 came closest by investigating 229 patients undergoing dobutamine echo followed by angiography, the largest study of contrast stress echo (for LVO) with angiography as the gold standard. As previously documented, the endocardial border definition and inter observer variability was superior with contrast. Their important finding was that there was comparable sensitivity, specificity and accuracy for the presence of coronary disease between patients with poor resting images using contrast LVO and those in whom standard imaging gave good resting images. Moreover, our recent work shows that the improvement of LVO is an important contributor to the incremental benefit of MCE (Figure 7 and additional files 10,11,12,13,14,15,16,17,18,19,20,21,22, and 23).

Thus, left ventricular opacification by contrast echo:

(1) improves the visualization of myocardial endocardial border definition;

(2) improves the accuracy of ventricular volume assessment and estimation of ejection fraction compared with standard fundamental and harmonic imaging.

(3) improves the ability to identify and grade resting wall motion abnormalities

(4) provides superior endocardial visualization while imaging at peak dobutamine stress and can enable the accuracy of a dobutamine stress echo in a technically difficult patient to be at least as good as that of a patient with good resting images;

(5) reduces the inter-observer variability for all of the above.

From an economic standpoint, the use of contrast agents during stress echo has been calculated to be cost effective 26 with the cost of the contrast agent itself more than offset by savings incurred by reducing downstream repetitive testing, improved laboratory efficiency and a lower rate of false positive and negatives. However, the calculations are based on a formulas incorporating improved accuracy of the technique for the diagnosis of CAD, compared with standard imaging, which remains un-proven.

In summary, the use of contrast for LVO is justified for standard or stress imaging of technically difficult patients, and possibly, for calculation of ventricular dimensions in patients whom accurate quantitative serial follow up is critical, eg. chemotherapy or valvular heart disease. Other clinical uses of contrast for left ventricular opacification include confirming or excluding the presence of left ventricular thrombus and delineating other left ventricular structures like pseudoaneurysm, apical hypertrophic cardiomyopathy (Figure 8 and additional files 24,25), and non-compacted left ventricles 272829.

In the ischaemic cascade, hypoperfusion precedes wall motion abnormalities, which precede ECG changes and the onset of chest discomfort 30. Stress echo, which provides an indirect marker of hypo-perfusion by recognition of wall motion abnormalities, is more than 80% sensitive and specific for detection of CAD 20. The technique has two groups of limitations;

1) Interpretive. Wall motion analysis is fundamentally subjective 31, leading to the need for specific training and discordance between observers. There are particular problems in the assessment of the inferoposterior wall.

2) Dependence on the induction of ischaemia (increased myocardial oxygen demand) to provide diagnostic information. Many problems spring from this – dependence on performance of adequate stress, influence by beta blockade, limited sensitivity for mild (eg single vessel) CAD, poor capacity to identify the extent of CAD (eg. 50% sensitivity for the detection of multi vessel disease)32.

SPECT perfusion imaging is the best-established clinical method to address perfusion directly and theoretically should be more sensitive than stress echo because hypoperfusion occurs earliest the ischaemic cascade. However, the difference between the methods is marginal because of the technical limitations of SPECT scanning, which also relies on subjective interpretation.

If technically feasible there is clear role for echocardiography to assess myocardial perfusion. Ultrasound is more widely available, portable, avoids radiation and has a better spatial resolution than SPECT scanning (up to 1 mm, compared with >10 mm). An ideal technique would allow assessment of myocardial perfusion and wall motion simultaneously.

Understanding the pathophysiology of normal myocardial perfusion is essential for understanding detection of CAD with MCE. Contrast micro-bubbles act as pure intravascular tracers, traversing the myocardial vasculature, and contrasting with MRI tracers (which escape into the extravascular space) and radionuclides (which enter myocytes).

There are three aspects to myocardial contrast opacification – signal intensity (equivalent to myocardial blood volume), pattern of filling and rate of filling.

Qualitative assessment of perfusion with MCE was validated in open chest dogs using graded coronary stenosis 37. The original landmark paper in humans 8 compared MCE (intermittent harmonic MCE with offline digital subtraction and colour coding) with sestamibi SPECT scanning in 30 patients undergoing dipyridamole stress to investigate known or suspected coronary disease. Normal segments showed a 91% concordance, and abnormal segments an 85% concordance, with 90% concordance by vascular territory. There was an overall 86% concordance for the detection of coronary disease. Subsequent studies have validated MCE following vasodilator or dobutamine stress using other imaging modalities, including power Doppler and real time power pulse inversion/power modulation in animals and in humans have produced similar results to SPECT 383940. A recent study using real time imaging was the first to report the use of MCE with exercise stress 41 (which is technically challenging because of cardiorespiratory movement and short duration of hyperaemia) and again demonstrated concordance between MCE and SPECT imaging. There was a 76% agreement between MCE and SPECT and an 88% agreement between the combination of wall motion and MCE with SPECT. This study incorporated the wall motion data from real time imaging, and despite the low frame rate, the results hint at the incremental benefit of combining the approaches.

While numerous studies examining various stress and imaging modalities have demonstrated concordance between MCE and SPECT, there remains a paucity of data using quantitative coronary angiography (QCA) as the gold standard. The presence and extent of CAD by QCA has traditionally been used as the reference standard for the assessment of CAD, but few studies have compared MCE and angiography (Table 3) 42. In 45 patients undergoing mainly dobutamine stress echocardiography, Cwajg et al showed real time contrast-enhanced imaging was more sensitive than standard stress echocardiography (87% vs 56%), and gave a better recognition of disease extent (85% vs 39%). These results were surprisingly unfavourable for standard imaging, perhaps reflecting the limitations of lower frame-rate for real-time imaging. Moreover, no specificity data were available from this study. In the study of exercise stress (41), only 44 of the 100 patients who had MCE and SPECT proceeded to angiography, and in this group, the sensitivity and specificity of wall motion and MCE were not significantly different. In a recent study of 85 prospectively recruited patients, the largest involving quantitative coronary angiography, we demonstrated the addition of contrast (for LVO and MCE), to standard ExE significantly enhanced the sensitivity of the test for detection of CAD from 74 to 91%, with a non-significant reduction in specificity (57).

Like stress echo and nuclear perfusion imaging, MCE is also limited by the qualitative nature of the interpretation. Subtle differences in video intensity between vascular beds may not be visually evident, potentially reducing the sensitivity for the detection of stenosis. Quantitative methods may help to alleviate this limitation and possibly reduce intra-observer and inter-observer variability in assessment.

The process of quantitative myocardial contrast echo was validated in open chest dogs using intermittent imaging, and the same destruction-replenishment approach is used for contemporary qualitative assessment 43. At a steady state during continuous intravenous infusion of micro bubbles, the number of bubbles entering or leaving any capillary unit is constant and depends on the flow rate of the bubbles. If the micro bubbles are destroyed at time zero, the video intensity in a selected myocardial region is close to zero dB (black). With time, bubbles will replenish the beam elevation – the degree of replenishment into the beam elevation increases as the time after destruction is increased until eventually the entire ultrasound beam elevation is replenished and a plateau is reached, whereby no further increase in video-intensity/brightness can occur (Figures 15 and 16). Because this relationship was originally described with intermittent imaging, it is usually described in terms of video-intensity vs pulsing intervals.

The video intensity vs pulsing interval or time curve resembles an exponential growth function that can be fitted by the equation I = A (1-e-βt) where y is the video intensity at the pulsing interval or time t, A is the plateau video intensity and β represents the rate of rise of the slope. In this model the plateau myocardial video intensity (A) represents the myocardial blood volume. The rate of rise of the slope (β) represents the mean myocardial red blood cell velocity and their product (A*β) represents myocardial blood flow. Wei demonstrated an excellent linear relationship between absolute myocardial blood flow measured with radiolabelled microspheres and myocardial contrast echo derived myocardial blood flow. While ECG gated intermittent triggered harmonic imaging was used to validate these measurements, both power Doppler and real time techniques have subsequently demonstrated similar results in animal experiments (444546).

More recently, Wei evaluated the ability of MCE to calculate flow reserves from these measurements in humans (47), using intermittent imaging in 30 selected patients undergoing coronary angiography (11 of whom had no CAD and 19 had non-critical single vessel stenoses). Quantitative MCE and invasive measurement coronary flow (Doppler flow wire) were performed at rest and following vasodilator stress. In the normal subjects, myocardial blood flow velocity (β) and myocardial blood flow (A*β) reserves demonstrated a linear relationship to coronary blood flow reserve measured invasively. In patients with CAD, there were significant differences in MBF velocity reserves between patients with mild moderate and severe stenoses, and a MBF velocity reserve of < 1.8 indicated a >70% stenosis. Thus MBF reserve appears to be a feasible non-invasive measure of CBF reserve in humans, which may allow non-invasive assessment of CAD and micro-vascular dysfunction. More recently, Dawson investigated the use of quantitatively derived MBF reserves to diagnose CAD using SPECT as the gold standard. She demonstrated moderate feasibility, with quantitative MBF reserves from both high and low MI imaging able to identify perfusion defects (48). Low MI imaging had a lower sensitivity.

It is now well recognised that regional or global ventricular dysfunction does not necessarily imply irreversible necrosis. Hypokinetic, akinetic or dyskinetic, yet viable myocardium may be stunned or hibernating. Myocardial stunning occurs after a period of acute ischaemia, despite restoration of completely normal blood flow. The natural history of stunning is of spontaneous improvement in the viable myocardium over time. Hibernating myocardium is the term used to describe the presence of significant ventricular dysfunction in patients with chronic CAD, which recovers after revascularisation. Improvement in function of sufficient numbers of viable but hibernating segments is associated with symptomatic benefit and improved survival49. Sadly, many patients with chronic CAD and LV dysfunction have minimal viability, and revascularisation of these patients is associated with significant risk, minimal benefit and possibly worse outcome, hence the need for a reliable test for identification of viability.

Radionuclide scanning, dobutamine echocardiography, MRI and PET scanning are currently available modalities, each with various advantages and disadvantages but similar efficacy for prediction of myocardial functional recovery after revascularization. MCE may also have an important role in this clinical setting. Viable myocardium is associated with preservation of the micro-vasculature, and as micro-bubbles act as pure intra-vascular tracers, the presence of myocardial perfusion by any MCE technique at rest implies viability (see figure 17). Using intra-coronary injection of bubbles, Nagueh demonstrated MCE was feasible and had similar accuracy to thallium SPECT and dobutamine echo for identification of functional recovery50. Using intravenous micro-bubble administration, MCE demonstrated comparable efficacy to SPECT and DSE. Importantly in both of these studies quantitative MCE was superior to qualitative assessment51. Unfortunately there remains a paucity of further data in this clinically important area.

The incidence of reported adverse events in human trials (mainly investigating LV opacification) has been very low. There have been particular theoretical concerns raised about histologic abnormalities and cardiac marker elevation in animals with high MI imaging52, and whilst serum tropinin levels are normal in humans after high MI imaging53, recent work has demonstrated troponin I and myoglobin in coronary sinus samples of humans after high MI imaging54.

In recent months the agent SonoVue (a phospho-lipid shell encapsulating sulphur hexafluoride gas), approved in Europe, was withdrawn from clinical use because of adverse events including fatalities related to idiosyncratic hypersensitivity reactions. At this stage, specific details about the dose of agent and imaging modality used in these cases is not available.