Danielewska, Monika E.*; Iskander, D. Robert†; Krzyżanowska-Berkowska, Patrycja‡
Corneal expansion, defined as the ocular pulse, is primarily caused by choroidal blood volume pulsation and the natural variations in the intraocular pressure (IOP).1–3 The IOP pulse is produced by the pulsatile arterial inflow and steady venous outflow.4 The ocular pulse is closely related to the heart activity.5–7 Recent results, based on phase analysis, have indicated that the electrical signal of the heart activity might have an independent role in axial corneal displacement.8 Variations in ocular blood supply are often reflected in changes of ocular pulse parameters, such as its amplitude, shape, and duration.1,9 Early works have widely investigated the ocular pulse amplitude as a sensitive indicator of some arteriosclerotic diseases.1,2,10–13 Less attention, however, has been given to the description and study of the ocular pulse characteristics.
In 1971, a study of the shape of the corneal indentation pulse (CIP), defined as temporal variations in corneal expansion, showed that pathological alterations in ocular blood supply are associated with corresponding changes in the form of the CIP signal.14 Furthermore, values of pulsatile ocular blood flow have proved to be strongly influenced by age, particularly in subjects older than 50 years.15 Aging is an important factor influencing changes in ocular rigidity, which, in turn, reflects changes in the elastic properties of the eye globe.16 It was demonstrated that, with aging, alterations of the collagen framework in corneal stroma occur.17,18 Results obtained by Pallikaris et al. showed a statistically significant positive correlation between the ocular rigidity coefficient and age.19 Also, there is evidence that ocular rigidity has particular relevance in pathologic myopia, macular degeneration, and glaucoma.19–22 Thus, a shape analysis of CIP and its aged-related interactions with cardiovascular signal parameters might gain more insight to hemodynamic aspects of ocular vascular diseases.
The main purpose of this work was to evaluate age-related changes in the averaged shape of CIP in relation to blood pulsation (BPL) and electrocardiogram (ECG) signals in healthy subjects.
Measurements were obtained from 77 healthy eyes in two age groups: 36 young subjects up to 35 years (mean ± SD, 28.3 ± 2.6 years; range, 23 to 32 years) and 41 older subjects older than 35 years (mean ± SD, 59.2 ± 7.0 years; range, 44 to 72 years). All subjects underwent general medical history review and standard ophthalmological examination to ensure that they had no reported ocular or cardiovascular abnormalities, were not regular contact lens wearers, and were free of any systemic diseases. Before CIP measurement, the subject’s IOP was measured with the Reichert 7CR auto-tonometer (Reichert, Inc., Depew, NY), whereas systolic and diastolic pressures were measured with the Blood Pressure Monitor HEM 780 (Omron Healthcare Co., Ltd., Japan). All subjects exhibited normal levels of IOP (mean ± SD, 13.9 ± 3.0 mm Hg) and normal levels of systolic blood pressure (range, 115 to 127 mm Hg) and diastolic pressure (range, 76 to 85 mm Hg). Exclusion criteria included intraocular surgery, refractive surgery, conjunctival or intraocular inflammation, and corneal abnormalities, such as edema or scars, the presence of which was affirmed by an ophthalmologist during a slit lamp examination.
The experiment was approved by the ethics committee of the Wrocław Medical University (KB 503/2011) and adhered to the tenets of the Declaration of Helsinki. The study protocol was explained and a signed informed consent was obtained from participating subjects before testing.
Subjects were sitting in a relaxed position and instructed to breathe naturally (∼0.2 ± 0.3 Hz).23 Measurement of the CIP signal requires adequately sensitive instrumentation.14 For this, we used an innovative ultrasonic distance sensor developed by UltraLab, Wroclaw, Poland, which allows noninvasive recording of the CIP amplitude with accuracy less than 1 μm.24 The transducer, operating at a frequency of 0.8 MHz,24 was placed in the working space of the transducer approximately 12 to 15 mm in front of the left cornea. In addition, to increase the precision of CIP measurements, a heavy headrest was applied to stabilize and reduce head movements.25,26 Simultaneously, continuous measurements of BPL and ECG signals were made. The BPL signal, which reflects oxygen saturation in the blood during each heartbeat, was recorded with a pulse oximeter placed on the right earlobe of a subject. Electrocardiogram signal measurements were taken in a standard three-lead system (Einthoven triangle).27 The sampling frequency for all considered signals was set to 400 Hz. Three 10-second recordings were obtained for each subject. During that time, the subject was asked to abstain from blinking and fixate on a designated stationary target set at his or her far point. During the measurement, a real-time preview of the received echo signal is available. This allows monitoring the potential blinks and interference resulting from substantial eye movements. The first recording containing no blinks and no eye movement interference was saved for further analysis.
Numerical processing of the CIP, BPL, and ECG signals was performed in a custom program written in MATLAB (MathWorks, Inc., Natick, MA). Initially, to remove the respiration related modulation, as well as to broaden the QRS complex of the ECG signal (corresponding to the heart ventricles depolarization), all analyzed signals were detrended and filtered from 0.5 to 20 Hz using a linear phase filter.
Classical signal averaging methods to analyze the shape of CIP, BPL, and ECG signals proved to be inadequate. All considered signals are nonstationary in nature—meaning that their frequency spectra vary in time.7 This is caused mainly by the heart rate variability and respiratory sinus arrhythmia.28 In consequence, a time-warping problem appears in subsequent cardiac cycles of these signals. A way to resolve this problem was to apply the algorithm of Dynamic Time Warping (DTW).29,30 Dynamic Time Warping finds an optimal alignment between two given time-dependent sequences through their “warped” (stretched and compressed) nonlinearity. Originally, DTW algorithm has been devised for speech recognition.31 The DTW algorithm was applied to obtain average warped shapes of CIP, BPL, and ECG signals for one cardiac cycle.
First, the principal period for each 10-second ECG measurement was calculated in the following way. Time positions of R peaks of the ECG signal were detected and marked to compute the durations of subsequent heartbeat cycles.32 The R peak is the central and usually the highest deflection in the QRS complex in a typical ECG and corresponds to the depolarization of the right and left ventricles of the human heart. Next, the average heartbeat cycle duration was calculated from the given ECG record. The cycle that was closest to the calculated average cycle was selected for further analysis as the principal one. Corresponding principal periods were extracted from the CIP and BPL signals. Finally, the DTW algorithm was used to compute averaged CIP, BPL, and ECG shapes from their whole records, with reference to the principal periods.
The presented procedure was performed for all measurements in this study. Fig. 1 shows typical time representations of simultaneously registered CIP, BPL, and ECG signals for a young subject as well as the results of their shape averaging using the DTW algorithm.
Based on the averaged signal shapes, the following parameters were computed (see panels A and B of Fig. 2, illustrating a typical measurement of a young and an older subject, respectively):
* time delay between systolic BPL peak and CIP maximum, τ(BPL, maxCIP);
* time delay between R peak of ECG and CIP maximum, τ(ECG, maxCIP);
* crest time (CT) (time taken from minimum to maximum of CIP shape);
* time duration of heart cycle (HCT) (taken from R peak to the next R peak of ECG shape); and
* relative crest time (RCT) = CT/HCT * 100%.
In subjects of the older group, some of whom exhibited a double peak–shaped CIP waveform, in addition to CT, HCT, and RCT, the following parameters were considered:
* time delay between systolic BPL peak and the first major CIP maximum, τ(BPL, maxCIP1);
* time delay between systolic BPL peak and the second major CIP maximum, τ(BPL, maxCIP2);
* time delay between R wave peak of ECG and the first major CIP maximum, τ(ECG, maxCIP1);
* time delay between R wave peak of ECG and the second major CIP maximum, τ(ECG, maxCIP2);
* preliminary crest time (preCT) (time taken from minimum to the first major maximum of CIP shape); and
* preliminary relative crest time (preRCT) = preCT/HCT * 100%.
In the absence of a double peak–shaped CIP waveform, the parameters τ(BPL, maxCIP2) and τ(ECG, maxCIP2) are equivalent to τ(BPL, maxCIP) and τ(ECG, maxCIP), respectively. A double-peak waveform has been assessed based on the relative relationship between the peaks in the CIP signal. A −3 dB criterion was used. That is, a double-peak waveform was declared when the valley between any two peaks in a signal (in the R-to-R range) was below a level corresponding to 0.707 of the lower of the two peaks.
All young subjects showed no evidence of double peak–shaped ocular pulsation. In the older group, 29 of 41 subjects exhibited this phenomenon. Correlation analysis based on orthogonal linear regression was used to identify the possible interactions between the shape parameters of the averaged signals for all subject measurements and separately for the two subject groups (young and older). Fig. 3 shows the correlation results between the considered shape parameters (CT, preCT, τ[BPL, maxCIP] = τ[BPL, maxCIP2], τ[ECG, maxCIP] = τ[ECG, maxCIP2], τ[BPL, maxCIP1], τ[ECG, maxCIP1], and HCT) obtained for young and older subjects. Note that the preCT parameter needs to be related to other parameters of the first peak (denoted by subscript 1) whereas the CT to those of the second peak (denoted by a subscript 2).
For the group of young subjects, where four signal shape parameters were considered (i.e., CT, τ[BPL, maxCIP], τ[ECG, maxCIP], and HCT), significant correlations (p < 0.05) were found between CT and τ(BPL, maxCIP) (R = 0.88, p < 0.001), CT and τ(ECG, maxCIP) (R = 0.93, p < 0.001), and CT and HCT (R = 0.53, p < 0.001).
For the group of older subjects, where up to seven signal shape parameters were considered (i.e., preCT, CT, τ[BPL, maxCIP1], τ[BPL, maxCIP2], τ[ECG, maxCIP1], τ[ECG, maxCIP2], and HCT), significant correlations were found between preCT and τ(BPL, maxCIP1) (R = 0.42, p = 0.002), preCT and τ(ECG, maxCIP1) (R = 0.78, p < 0.001), and preCT and HCT (R = 0.55, p = 0.002), but not for the corresponding correlations of CT.
Fig. 4 shows the RCT and preRCT values obtained for young and older subjects. For comparison, the range of RCT values obtained by Hørven and Nornes14 (from 35.5 to 47.5%) was marked by the dashed lines.
In earlier studies,8 evaluation of time delays between the ocular pulse and cardiovascular activity signals has been indicated to be a difficult task. Applying classical tools to the analysis of their interdependencies is often ambiguous because of the nonstationary nature of these signals. A new way to estimate time delays between CIP, BPL, and ECG signals was proposed. The dynamic time-warping algorithm has been applied to average their shapes and help resolve the time-warping problem for analyzed nonstationary signals.
In a group of young subjects and about 30% (12 of 41) of the older subjects, the maximum of the averaged CIP shape, which corresponds to the mechanical response to the blood pressure wave, follows the BPL peak and the R peak of ECG. This is an expected result. However, for the remaining older subjects, a clear evidence of a double peak–shaped character of the CIP signal was found. Smith and Craige33 found such a characteristic M shape waveform in the context of dicrotic pulse. In those cases, there were two major maxima of the averaged CIP shape, the first appearing before the BPL peak and the second occurring after the BPL peak.
For the group of young subjects, an increase in the crest time CT corresponded to an increase in the time delay between CIP maximum and BPL peak, τ(BPL, maxCIP), time delay between R peak of ECG and CIP maximum τ(ECG, maxCIP), as well as the time duration of the HCT (see the top row of Fig. 3). Similar relationships have been observed for the older group of subjects but for the preliminary crest time preCT (middle row of Fig. 3), with the difference that a decrease in the time advance (i.e., negative τ[BPL, maxCIP1]) between CIP maximum and BPL peak was observed with an increased CT. Lack of significant correlation between CT and the time delay parameters for the older group of subjects has currently no full explanation but might be attributed to the variations in the notch position of the M-shaped CIP signal.
A natural question arises whether the CIP signal changes observed in the substantial portion of older subjects (ca. 70%) can be viewed as the ocular pulse dicrotism. The characteristic M shape of the CIP waveform could be associated with age-progressive changes in blood wave propagation to the interior of the eye. It is known that an increase in arterial stiffness and pulse wave velocity occurs with aging.34,35 This leads to changes in the arterial pulse wave contour, evidenced in various body locations.36–40 It was also reported that older subjects demonstrate progressively earlier pressure wave reflection.36 Mac-Way et al. have explained how the timing of peak reflected pressure wave depends on the arterial stiffness.41
The principal character of a dicrotic waveform is the presence of two distinct peaks. The mechanism of the dicrotic pulse has been studied from the late 19th century.33,42–44
The pulse dicrotism has been interpreted as a distortion of the waveform resulting from abnormal arterial resistance45 and some heart diseases.46,47 However, there have been no reports related to the presence of dicrotism in ocular circulation. Because rigidity of the arterial walls tends to increase dicrotism,48 this process might also have an effect on the characteristics of ocular blood pressure wave and eye globe expansion in relation to age.
We speculate that the observed dicrotic ocular pulse in the older group of subjects could reflect higher arterial stiffness,35 higher ocular rigidity,19 and changes in blood flow in ocular vascular system with age.15 Worth noting is that the subgroup average age of older subjects showing ocular dicrotism was higher to that not exhibiting it (61 years vs. 55 years, respectively).
To the best of our knowledge, this is the first time when the ocular pulse dicrotism has been observed. Whether this is a natural sign of aging or an early indication of hemodynamic aspects of ocular vascular diseases is still a matter of further studies. In the latter case, noninvasive measurement of corneal indentation pulse would prove to be a window to early diagnosis of cardiovascular diseases. In the future, it would be also of interest to find out whether the ocular pulse dicrotism affects the visual function of the elderly subjects.
Monika E. Danielewska
Wybrzeze Wyspianskiego 27
Supported in part by the National Science Centre grant number UMO-2011/01/N/ST7/05391 and funds from the Foundation for Polish Science (project VENTURES/2011-7/4 to MED).
Monika E. Danielewska had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. None of the authors has a financial or proprietary interest in any material nor method mentioned nor received a financial support for this study from a commercial source. Part of this work was presented at the Association for Research in Vision and Ophthalmology Annual Meeting 2013 in Seattle.
Received May 13, 2013; accepted August 22, 2013.
1. Zuckerman JL, Taylor KD, Grossman HJ. Noncontact detection of ocular pulse—correlation with carotid stenosis. Invest Ophthalmol Vis Sci 1977; 16: 1018–24.
2. Northrop RB, Nilakhe SS. A no-touch ocular pulse measurement system for the diagnosis of carotid occlusions. IEEE Trans Biomed Eng 1977; 24: 139–48.
3. Langham ME, Farrell RA, O’Brien V, Silver DM, Schilder P. Blood flow in the human eye. Acta Ophthalmol Suppl 1989; 191: 9–13.
4. Kiel JW. The Ocular Circulation. San Rafael, CA: Morgan & Claypool Life Sciences; 2010.
5. Bosley TM, Cohen MS, Gee W, Reed J, 3rd, Sergott RC, Savino PJ. Amplitude of the ocular pneumoplethysmography waveform is correlated with cardiac output. Stroke 1993; 24: 6–9.
6. Trew DR, James CB, Thomas SH, Sutton R, Smith SE. Factors influencing the ocular pulse—the heart rate. Graefes Arch Clin Exp Ophthalmol 1991; 229: 553–6.
7. Kasprzak HT, Iskander DR. Spectral characteristics of longitudinal corneal apex velocities and their relation to the cardiopulmonary system. Eye (Lond) 2007; 21: 1212–9.
8. Danielewska ME, Iskander DR, Kowalska M, Kasprzak HT. Phase dependencies between longitudinal corneal apex displacement and cardiovascular signals: is the ocular pulse influenced by the electrical activity of the heart? Clin Exp Optom 2012; 95: 631–7.
9. Best M, Pola R, Plechaty G, Masket S, Galin MA. Ocular pulse studies in carotid stenosis. Relationship to carotid hemodynamics. Arch Ophthalmol 1971; 85: 730–7.
10. Best M, Keyes M, Plechaty G, Galin MA. Graphic analysis of the ocular pulse in carotid occlusion. Arch Ophthalmol 1971; 85: 315–9.
11. Schilder P. Ocular blood flow responses to pathology of the carotid and cerebral circulations. Surv Ophthalmol 1994; 38 (Suppl.): S52–8.
12. Schilder P. Ocular blood flow changes with increased vascular resistance external and internal to the eye. Acta Ophthalmol Suppl 1989; 191: 19–23.
13. Costa VP, Kuzniec S, Molnar LJ, Cerri GG, Puech-Leão P, Carvalho CA. The effects of carotid endarterectomy on the retrobulbar circulation of patients with severe occlusive carotid artery disease. An investigation by color Doppler imaging. Ophthalmology 1999; 106: 306–10.
14. Hørven I, Nornes H. Crest time evaluation of corneal indentation pulse. A new parameter in the study of ocular blood supply. Arch Ophthalmol 1971; 86: 5–11.
15. Ravalico G, Toffoli G, Pastori G, Crocè M, Calderini S. Age-related ocular blood flow changes. Invest Ophthalmol Vis Sci 1996; 37: 2645–50.
16. Friedenwald JS. Contribution to the theory and practice of tonometry. Am J Ophthalmol 1937; 20: 985–1024.
17. Malik NS, Moss SJ, Ahmed N, Furth AJ, Wall RS, Meek KM. Ageing of the human corneal stroma: structural and biochemical changes. Biochim Biophys Acta 1992; 1138: 222–8.
18. Daxer A, Misof K, Grabner B, Ettl A, Fratzl P. Collagen fibrils in the human corneal stroma: structure and aging. Invest Ophthalmol Vis Sci 1998; 39: 644–8.
19. Pallikaris IG, Kymionis GD, Ginis HS, Kounis GA, Tsilimbaris MK. Ocular rigidity in living human eyes. Invest Ophthalmol Vis Sci 2005; 46: 409–14.
20. Baker BJ, Pruett RC. Degenerative myopia. In: Yanoff M, Duker JS, eds. Ophthalmology, 3rd ed. Edinburgh, UK: Mosby Elsevier; 2009: 674–6.
21. Friedman E. The role of the atherosclerotic process in the pathogenesis of age-related macular degeneration. Am J Ophthalmol 2000; 130: 658–63.
22. Hommer A, Fuchsjager-Mayrl G, Resch H, Vass C, Garhofer G, Schmetterer L. Estimation of ocular rigidity based on measurement of pulse amplitude using pneumotonometry and fundus pulse using laser interferometry in glaucoma. Invest Ophthalmol Vis Sci 2008; 49: 4046–50.
23. Sherwood L. Fundamentals of Physiology: A Human Perspective, 3rd ed. Belmont, CA: Brooks/Cole; 2006.
24. Licznerski TJ, Jaronski J, Kosz D. Ultrasonic system for accurate distance measurement in the air. Ultrasonics 2011; 51: 960–5.
25. Kasprzak HT, Iskander DR. Ultrasonic measurement of fine head movements in a standard ophthalmic headrest. IEEE Trans Instr Meas 2010; 59: 164–70.
26. Kowalska MA, Kasprzak HT, Iskander DR, Danielewska M, Mas D. Ultrasonic in vivo measurement of ocular surface expansion. IEEE Trans Biomed Eng 2011; 58: 674–80.
27. Conover MB. Electrocardiography, 5th ed. St. Louis, MO: Mosby; 2004.
28. Angelone A, Coulter NA. Respiratory sinus arrhythmia: a frequency dependent phenomenon. J Appl Physiol 1964; 19: 479–82.
29. Sakoe H, Chiba S. Dynamic programming algorithm optimization for spoken word recognition. IEEE Trans Acoust Speech 1978; 26: 43–9.
30. Boudaoud S, Rix H, Meste O. Integral shape averaging and structural average estimation: a comparative study. IEEE Trans Signal Process 2005; 53: 3644–50.
31. Rabiner L, Rosenberg AE, Levinson SE. Considerations in dynamic time warping algorithms for discrete word recognition. IEEE Trans Acoust Speech 1978; 26: 575–82.
32. Pan J, Tompkins WJ. A real-time QRS detection algorithm. IEEE Trans Biomed Eng 1985; 32: 230–6.
33. Smith D, Craige E. Mechanism of the dicrotic pulse. Br Heart J 1986; 56: 531–4.
34. Lakatta EG, Mitchell JH, Pomerance A, Rowe GG. Human aging: changes in structure and function. J Am Coll Cardiol 1987; 10: 42A–7A.
35. Vaitkevicius PV, Fleg JL, Engel JH, O’Connor FC, Wright JG, Lakatta LE, Yin FC, Lakatta EG. Effects of age and aerobic capacity on arterial stiffness in healthy adults. Circulation 1993; 88: 1456–62.
36. Kelly R, Hayward C, Avolio A, O’Rourke M. Noninvasive determination of age-related changes in the human arterial pulse. Circulation 1989; 80: 1652–9.
37. Allen J, Murray A. Age-related changes in the characteristics of the photoplethysmographic pulse shape at various body sites. Physiol Meas 2003; 24: 297–307.
38. Sherebrin MH, Sherebrin RZ. Frequency analysis of the peripheral pulse wave detected in the finger with a photoplethysmograph. IEEE Trans Biomed Eng 1990; 37: 313–7.
39. Oliva I, Ipser J, Roztocil K, Guttenbergerova K. Fourier analysis of the pulse wave in obliterating arteriosclerosis. Vasa 1976; 5: 95–100.
40. Millasseau SC, Kelly RP, Ritter JM, Chowienczyk PJ. Determination of age-related increases in large artery stiffness by digital pulse contour analysis. Clin Sci (Lond) 2002; 103: 371–7.
41. Mac-Way F, Leboeuf A, Agharazii M. Arterial stiffness and dialysis calcium concentration. Int J Nephrol 2011; 2011: 839793.
42. Fleming WJ. Pulse dicrotism. J Anat Physiol 1881; 15: 278–91.
43. MacKenzie J. The Study of the Pulse; Arterial, Venous and Hepatic and of the Movement of the Heart. London, UK: Y. J. Pentland; 1902.
44. O’Rourke MF, Gallagher DE. Pulse wave analysis. J Hypertens Suppl 1996; 14: S147–57.
45. Orchard RC, Craige E. Dicrotic pulse after open heart surgery. Circulation 1980; 62: 1107–14.
46. Ewy GA, Rios JC, Marcus FI. The dicrotic arterial pulse. Circulation 1969; 39: 655–61.
47. Meadows WR, Draur RA, Osadjan CE. Dicrotism in heart disease. Correlations with cardiomyopathy, pericardial tamponade, youth, tachycardia, and normotension. Am Heart J 1971; 82: 596–608.
48. Lewis T. The factors influencing the prominence of the dicrotic wave. J Physiol 1906; 34: 414–29.