An Ejection Fraction Measurement for the Masses*

    loading  Checking for direct PDF access through Ovid

Excerpt

As far as we can tell, one of the very first descriptions of the left ventricular ejection fraction (LVEF) was published by Holt (1) from Louisville, Kentucky in 1956. “The ventricle empties itself in a ‘fractional’ manner, approximately 46 per cent of its end-diastolic volume being ejected with each stroke.…” That article—well-worth a read for the heroism of the experiments and the clarity of the thinking—describes both dye-dilution and electrical conductivity approaches, the former requiring blood drawn quantitatively from the canine aortic arch (ironically, this was the year that Needles won the Derby), and the latter requiring purpose-built conductivity cells inserted into the same structure.
Now comes Pahlevan et al (2), in this issue of Critical Care Medicine, who, looking through the eye of an iPhone (Apple, Cupertino, CA) and analyzing frequency domain characteristics (harmonics, perhaps, to Wordsworth) of the skin overlying the carotid artery, can see into the life of the heart and report LVEF, all in a matter of moments. The underlying principles—separate frequency characteristics of the heart and vasculature that are coupled during systole but uncoupled during diastole—are well-explained in an earlier publication (3), and the central analytic, the counter-intuitive notion that a part of a wave has frequency characteristics, are nonetheless well-justified. Determining the dominant intrinsic frequencies before and after the aortic valve closure is not only justified from a physiologic standpoint but also has attractive features built on mathematical principles. The sparse time-frequency representation method is inspired by the empirical mode decomposition (4) and intrinsic mode functions that are particularly well-suited to aortic pressure waves, which are locally well-modeled by signals with slowly varying single frequencies.
Under this framework, modeling the pulse as being piecewise constant in time before and after the dicrotic notch is a natural and elegant application. The use of a brute force algorithm to perform exhaustive searches and to avoid local extrema is a wise choice and computationally feasible. The accuracy of the model is demonstrated in the reconstruction of the four diverse examples in Figure 1 (2). The example in Figure 6 does a particularly good job of demonstrating how the instantaneous frequencies change after the dicrotic notch. Ultimately, the relative simplicity of the approach might widen its clinical applicability while maintaining a solid physiologic foundation.
This pilot clinical validation makes its point that the new measure can as reasonable as the standard practice of echocardiography. In this study, Figure 2A (2) makes it clear that there is better agreement of the ejection fraction (EF) from MRI and the new device for LVEF values of less than 40% than above, which may well be clinically acceptable as a first test for left ventricle (LV) dysfunction. Clinicians will be looking forward to much larger-scale studies under real-world conditions of patients with well-detailed heart disease, and studies of how the new measurements fit with clinical decision-making and workflows. It will be useful as well to see how the measure is affected by other common heart diseases such as aortic and mitral valve disease, and by medications and blood counts.
It is easy to think of how handy this might be. For the Critical Care readership, consider the triage potential of this device in evaluation of the patient with new-onset shock. For the oncologist, consider this part of the chemotherapy visit. For the general cardiologist, consider this as a new vital sign to be recorded at outpatient visits. For the cardiac electrophysiologist, consider this a screening tool for finding patients with low EF that are candidates for primary prevention implanted defibrillators.
    loading  Loading Related Articles