Measurement of normal retinal vascular pulse wave attenuation using modified photoplethysmography

Probability density and information entropy of machine learning derived intracranial pressure predictions

Even with the powerful statistical parameters derived from the Extreme Gradient Boost (XGB) algorithm, it would be advantageous to define the predicted accuracy to the level of a specific case, particularly when the model output is used to guide clinical decision-making. The probability density function (PDF) of the derived intracranial pressure predictions enables the computation of a definite integral around a point estimate, representing the event's probability within a range of values. Seven hold-out test cases used for the external validation of an XGB model underwent retinal vascular pulse and intracranial pressure measurement using modified photoplethysmography and lumbar puncture, respectively. The definite integral ±1 cm water from the median (DIICP) demonstrated a negative and highly significant correlation (-0.5213±0.17, p< 0.004) with the absolute difference between the measured and predicted median intracranial pressure (DiffICPmd). The concordance between the arterial and venous probability density functions was estimated using the two-sample Kolmogorov-Smirnov statistic, extending the distribution agreement across all data points. This parameter showed a statistically significant and positive correlation (0.4942±0.18, p< 0.001) with DiffICPmd. Two cautionary subset cases (Case 8 and Case 9), where disagreement was observed between measured and predicted intracranial pressure, were compared to the seven hold-out test cases. Arterial predictions from both cautionary subset cases converged on a uniform distribution in contrast to all other cases where distributions converged on either log-normal or closely related skewed distributions (gamma, logistic, beta). The mean±standard error of the arterial DIICP from cases 8 and 9 (3.83±0.56%) was lower compared to that of the hold-out test cases (14.14±1.07%) the between group difference was statistically significant (p<0.03). Although the sample size in this analysis was limited, these results support a dual and complementary analysis approach from independently derived retinal arterial and venous non-invasive intracranial pressure predictions. Results suggest that plotting the PDF and calculating the lower order moments, arterial DIICP, and the two sample Kolmogorov-Smirnov statistic may provide individualized predictive accuracy parameters.

Alterations in retinal pulse wave velocity under experimental ocular hypertension

Impairments of blood flow and autoregulation have been implicated in diabetic retinopathy and glaucoma. Thus, identifying biomarkers of retinal vascular compliance and regulatory capacity is of potential value for understanding the pathophysiology and evaluating onset or progression of disease. Pulse wave velocity (PWV) represents the speed of the pulse-propagated pressure wave within blood vessels and has shown promise as a marker of vascular compliance. The purpose of the current study was to report a method for comprehensive assessment of retinal PWV based on spectral analysis of pulsatile intravascular intensity waveforms and determine alterations due to experimental ocular hypertension. Retinal PWV was linearly related to vessel diameter. Increased retinal PWV was associated with elevated intraocular pressure. Retinal PWV has the potential to serve as a vasoregulation biomarker for investigating vascular factors that contribute to the development of retinal diseases in animal models.

Empirical retinal venous pulse wave velocity using modified photoplethysmography

OBJECTIVE

Using the novel imaging method of high-speed modified photoplethysmography we measured the retinal venous pulse wave velocity in a single case.

RESULTS

A healthy 30-year-old subject underwent high-speed modified photoplethysmography (120 frames per second) with simultaneous ophthalmodynamometry at 26 Meditron units. A video of the optic nerve was analyzed using custom software. A harmonic regression model was fitted to each pixel in the time series and used to quantify the retinal vascular pulse wave parameters. Retinal venous pulsation at the optic disc was observed as a complex dynamic wall motion, whereas contraction commenced at a point in the vein at the center of the optic disc, and progressed centrifugally. The empirically estimated retinal venous pulse wave velocity at this segment was approximately 22.24694 mm/s. This measurement provides an estimate for future studies in the field.

Correlation between retinal vein pulse amplitude, estimated intracranial pressure, and postural change

Spaceflight associated neuro-ocular syndrome (SANS) is common amongst astronauts on long duration space missions and is associated with signs consistent with elevated cerebrospinal fluid (CSF) pressure. Additionally, CSF pressure has been found to be elevated in a significant proportion of astronauts in whom lumbar puncture was performed after successful mission completion. We have developed a retinal photoplethysmographic technique to measure retinal vein pulsation amplitudes. This technique has enabled the development of a non-invasive CSF pressure measurement apparatus. We tested the system on healthy volunteers in the sitting and supine posture to mimic the range of tilt table extremes and estimated the induced CSF pressure change using measurements from the CSF hydrostatic indifferent point. We found a significant relationship between pulsation amplitude change and estimated CSF pressure change (p < 0.0001) across a range from 2.7 to 7.1 mmHg. The increase in pulse amplitude was highest in the sitting posture with greater estimated CSF pressure increase (p < 0.0001), in keeping with physiologically predicted CSF pressure response. This technique may be useful for non-invasive measurement of CSF pressure fluctuations during long-term space voyages.

Effect of intracranial pressure on photoplethysmographic waveform in different cerebral perfusion territories: A computational study

Background: Intracranial photoplethysmography (PPG) signals can be measured from extracranial sites using wearable sensors and may enable long-term non-invasive monitoring of intracranial pressure (ICP). However, it is still unknown if ICP changes can lead to waveform changes in intracranial PPG signals.

Aim: To investigate the effect of ICP changes on the waveform of intracranial PPG signals of different cerebral perfusion territories.

Methods: Based on lump-parameter Windkessel models, we developed a computational model consisting three interactive parts: cardiocerebral artery network, ICP model, and PPG model. We simulated ICP and PPG signals of three perfusion territories [anterior, middle, and posterior cerebral arteries (ACA, MCA, and PCA), all left side] in three ages (20, 40, and 60 years) and four intracranial capacitance conditions (normal, 20% decrease, 50% decrease, and 75% decrease). We calculated following PPG waveform features: maximum, minimum, mean, amplitude, min-to-max time, pulsatility index (PI), resistive index (RI), and max-to-mean ratio (MMR).

Results: The simulated mean ICPs in normal condition were in the normal range (8.87–11.35 mm Hg), with larger PPG fluctuations in older subject and ACA/PCA territories. When intracranial capacitance decreased, the mean ICP increased above normal threshold (&gt;20 mm Hg), with significant decreases in maximum, minimum, and mean; a minor decrease in amplitude; and no consistent change in min-to-max time, PI, RI, or MMR (maximal relative difference less than 2%) for PPG signals of all perfusion territories. There were significant effects of age and territory on all waveform features except age on mean.

Conclusion: ICP values could significantly change the value-relevant (maximum, minimum, and amplitude) waveform features of PPG signals measured from different cerebral perfusion territories, with negligible effect on shape-relevant features (min-to-max time, PI, RI, and MMR). Age and measurement site could also significantly influence intracranial PPG waveform.

Vessel Pulse Amplitude Mapping in Eyes With Central and Hemi Retinal Venous Occlusion

Purpose

The purpose of this study was to describe vessel pulse amplitude characteristics in eyes with central retinal vein occlusion (CRVO), hemiretinal vein occlusion (HVO), normal eyes (N1 N1), and the unaffected contralateral eyes of CRVO and HVO eyes (N1 CRVO and N1 HVO), as well as the unaffected hemivessels of HVO eyes (N2 HVO).

Methods

Ophthalmodynamometry estimates of blood column pulse amplitudes with modified photoplethysmography were timed against cardiac cycles. Harmonic analysis was performed on the vessel reflectance within 0.25 to 1 mm from the disc center to construct pulse amplitude maps. Linear mixed modeling was used to examine variable effects upon the log harmonic pulse amplitude.

Results

One hundred seven eyes were examined. Normal eyes had the highest mean venous pulse amplitude (2.08 ± 0.48 log u). CRVO had the lowest (0.99 ± 0.45 log u, P < 0.0001), followed by HVO (1.23 ± 0.46 log u, P = 0.0002) and N2 HVO (1.30 ± 0.59 log u, P = 0.0005). N1 CRVO (1.76 ± 0.34 log u, P = 0.52) and N1 HVO (1.33 ± 0.37 log u, P = 0.0101) had no significantly different mean amplitudes compared to N1 N1. Arterial amplitudes were lower than venous (P < 0.01) and reduced with venous occlusion (P < 0.01). Pulse amplitude versus amplitude over distance decreased along the N1 N1 vessels, with increasing slopes observed with CRVO (P < 0.01).

Conclusions

Pulse amplitude reduction and attenuation characteristics of arteries and veins in venous occlusion can be measured and are consistent with reduced vessel wall compliance and pulse wave transmission.

Translational Relevance

Retinal vascular pulse amplitudes can be measured, revealing occlusion induced changes, suggesting a role in evaluating the severity and progression of venous occlusion.

MSGANet-RAV: A multiscale guided attention network for artery-vein segmentation and classification from optic disc and retinal images

BACKGROUND

Retinal and optic disc images are used to assess changes in the retinal vasculature. These can be changes associated with diseases such as diabetic retinopathy and glaucoma or induced using ophthalmodynamometry to measure arterial and venous pressure. Key steps toward automating the assessment of these changes are the segmentation and classification of the veins and arteries. However, such segmentation and classification are still required to be manually labelled by experts. Such automated labelling is challenging because of the complex morphology, anatomical variations, alterations due to disease and scarcity of labelled data for algorithm development. We present a deep machine learning solution called the multiscale guided attention network for retinal artery and vein segmentation and classification (MSGANet-RAV).

METHODS

MSGANet-RAV was developed and tested on 383 colour clinical optic disc images from LEI-CENTRAL, constructed in-house and 40 colour fundus images from the AV-DRIVE public dataset. The datasets have a mean optic disc occupancy per image of 60.6% and 2.18%, respectively. MSGANet-RAV is a U-shaped encoder-decoder network, where the encoder extracts multiscale features, and the decoder includes a sequence of self-attention modules. The self-attention modules explore, guide and incorporate vessel-specific structural and contextual feature information to segment and classify central optic disc and retinal vessel pixels.

RESULTS

MSGANet-RAV achieved a pixel classification accuracy of 93.15%, sensitivity of 92.19%, and specificity of 94.13% on LEI-CENTRAL, outperforming several reference models. It similarly performed highly on AV-DRIVE with an accuracy, sensitivity and specificity of 95.48%, 93.59% and 97.27%, respectively.

CONCLUSION

The results show the efficacy of MSGANet-RAV for identifying central optic disc and retinal arteries and veins. The method can be used in automated systems designed to assess vascular changes in retinal and optic disc images quantitatively.

A machine learning approach in the non-invasive prediction of intracranial pressure using Modified Photoplethysmography

The ideal Intracranial pressure (ICP) estimation method should be accurate, reliable, cost-effective, compact, and associated with minimal morbidity/mortality. To this end several described non-invasive methods in ICP estimation have yielded promising results, however the reliability of these techniques have yet to supersede invasive methods of ICP measurement. Over several publications, we described a novel imaging method of Modified Photoplethysmography in the evaluation of the retinal vascular pulse parameters decomposed in the Fourier domain, which enables computationally efficient information filtering of the retinal vascular pulse wave. We applied this method in a population of 21 subjects undergoing lumbar puncture manometry. A regression model was derived by applying an Extreme Gradient Boost (XGB) machine learning algorithm using retinal vascular pulse harmonic regression waveform amplitude (HRWa), first and second harmonic cosine and sine coefficients (an1,2, bn1,2) among other features. Gain and SHapley Additive exPlanation (SHAP) values ranked feature importance in the model. Agreement between the predicted ICP mean, median and peak density with measured ICP was assessed using Bland-Altman bias±standard error. Feature gain of intraocular pressure (IOPi) (arterial = 0.6092, venous = 0.5476), and of the Fourier coefficients, an1 (arterial = 0.1000, venous = 0.1024) ranked highest in the XGB model for both vascular systems. The arterial model SHAP values demonstrated the importance of the laterality of the tested eye (1.2477), which was less prominent in the venous model (0.8710). External validation was achieved using seven hold-out test cases, where the median venous predicted ICP showed better agreement with measured ICP. Although the Bland-Altman bias from the venous model (0.034±1.8013 cm water (p<0.99)) was lower compared to that of the arterial model (0.139±1.6545 cm water (p<0.94)), the arterial model provided a potential avenue for internal validation of the prediction. This approach can potentially be integrated into a neurological clinical decision algorithm to evaluate the indication for lumbar puncture.

Near infra-red reflectance videography in the evaluation of retinal artery macroaneurysm pulsatility

Purpose

To describe pulsations of the retinal arteries detected in the course of evaluation of an exudative non-pulsatile retinal arterial macroaneurysm using near infra-red reflectance videography.

Observations

A 68-year-old patient underwent slit lamp examination, color retinal imaging, optical coherence tomography, fluorescein videography, short wave-length and near infrared fundus autofluorescence of the left, and near infrared reflectance videography of both eyes. A 1309.3 × 955.1 μm exudative lesion with intraretinal hemorrhage and retinal edema secondary to a retinal arterial macroaneurysm was observed along the superior temporal arcade between the retinal artery and vein. Bilateral serpentine and expansile spontaneous retinal artery pulsations were detected along the retinal vascular tree and imaged using near infrared reflectance videography. Fluorescein video-angiography showed an irregular filling defect of the lesion with minimal angiographic leakage. Whereas pulsations of the retinal arteries were visualized, no pulsations of the retinal arterial macroaneurysm were detected with either dynamic imaging modality, therefore observation was recommended. Significant spontaneous lesion regression was observed at one month follow-up.

Conclusionand Importance

Detection of spontaneous retinal artery pulsation and the assessment of exudative maculopathy due to an underlying retinal arterial macroaneurysm could be facilitated by near infrared reflectance videography. This imaging modality can aid in clinical decision-making where a non-pulsatile macroaneurysm would favor conservative management.

Linear interactions between intraocular, intracranial pressure, and retinal vascular pulse amplitude in the fourier domain

Purpose

To compare the retinal vascular pulsatile characteristics in subjects with normal (ICP_n) and high (ICP_h) intracranial pressure and quantify the interactions between intraocular pressure, intracranial pressure, and retinal vascular pulse amplitude in the Fourier domain.

Materials and methods

Twenty-one subjects were examined using modified photoplethysmography with simultaneous ophthalmodynamometry. A harmonic regression model was fitted to each pixel in the time-series, and used to quantify the retinal vascular pulse wave parameters including the harmonic regression wave amplitude (HRW_a). The pulse wave attenuation was measured under different ranges of induced intraocular pressure (IOP_i), as a function of distance along the vessel (V_Dist). Intracranial pressure (ICP) was measured using lumbar puncture. A linear mixed-effects model was used to estimate the correlations between the Yeo-Johnson transformed harmonic regression wave amplitude (HRW_a-YJt) with the predictors (IOP_i, V_Dist and ICP). A comparison of the model coefficients was done by calculating the weighted Beta (β_x) coefficients.

Results

The median HRW_a in the ICP_n group was higher in the retinal veins (4.563, interquartile range (IQR) = 3.656) compared to the retinal arteries (3.475, IQR = 2.458), p<0.0001. In contrast, the ICP_h group demonstrated a reduction in the median venous HRW_a (3.655, IQR = 3.223) and an elevation in the median arterial HRW_a (3.616, IQR = 2.715), p<0.0001. Interactions of the pulsation amplitude with ICP showed a significant disordinal interaction and the loss of a main effect of the Fourier sine coefficient (b_n1) in the ICP_h group, suggesting that this coefficient reflects the retinal vascular response to ICP wave. The linear mixed-effects model (LME) showed the decay in the venous (HRW_a-YJt) was almost twice that in the retinal arteries (−0.067±0.002 compared to −0.028±0.0021 respectively, p<0.00001). The overall interaction models had a total explanatory power of (conditional R²) 38.7%, and 42% of which the fixed effects explained 8.8%, and 5.8% of the variance (marginal R²) for the venous and arterial models respectively. A comparison of the damping effect of V_Dist and ICP showed that ICP had less influence on pulse decay than distance in the retinal arteries (β_ICP = -0.21, se = ±0.017 compared to βVDist=-0.26, se = ±0.019), whereas the mean value was equal for the retinal veins (venous βVDist=-0.42, se = ±0.015, β_ICP = -0.42, se = ±0.019).

Conclusion

The retinal vascular pulsation characteristics in the ICP_h group showed high retinal arterial and low venous pulsation amplitudes. Interactions between retinal vascular pulsation amplitude and ICP suggest that the Fourier sine coefficient b_n1 reflects the retinal vascular response to the ICP wave. Although a matrix of regression lines showed high linear characteristics, the low model explanatory power precludes its use as a predictor of ICP. These results may guide future predictive modelling in non-invasive estimation of ICP using modified photoplethysmography.

Zero retinal vein pulsation amplitude extrapolated model in non-invasive intracranial pressure estimation

Intracranial pressure (ICP) includes the brain, optic nerve, and spinal cord pressures; it influences blood flow to those structures. Pathological elevation in ICP results in structural damage through various mechanisms, which adversely affects outcomes in traumatic brain injury and stroke. Currently, invasive procedures which tap directly into the cerebrospinal fluid are required to measure this pressure. Recent fluidic engineering modelling analogous to the ocular vascular flow suggests that retinal venous pulse amplitudes are predictably influenced by downstream pressures, suggesting that ICP could be estimated by analysing this pulse signal. We used this modelling theory and our photoplethysmographic (PPG) retinal venous pulse amplitude measurement system to measure amplitudes in 30 subjects undergoing invasive ICP measurements by lumbar puncture (LP) or external ventricular drain (EVD). We estimated ICP from these amplitudes using this modelling and found it to be accurate with a mean absolute error of 3.0 mmHg and a slope of 1.00 (r = 0.91). Ninety-four percent of differences between the PPG and invasive method were between − 5.5 and + 4.0 mmHg, which compares favourably to comparisons between LP and EVD. This type of modelling may be useful for understanding retinal vessel pulsatile fluid dynamics and may provide a method for non-invasive ICP measurement.

Retinal venous pressure is decreased after anti-VEGF therapy in patients with retinal vein occlusion-related macular edema

Purpose

The pathomechanism leading to retinal vein occlusion (RVO) is unclear. Mechanical compression, thrombosis, and functional contractions of veins are discussed as the reasons for the increased resistance of venous outflow. We evaluated changes in the retinal venous pressure (RVP) following intravitreal injection of anti-vascular endothelial growth factor (VEGF) agent to determine the effect on RVO-related macular edema.

Methods

Twenty-six patients with RVO-related macular edema (16 branch RVOs [BRVOs] and 10 central RVOs [CRVOs], age 72.5 ± 8.8 years) who visited our hospital were included in this prospective study. Visual acuity (VA), intraocular pressure (IOP), central retinal thickness (CRT) determined by macular optical coherence tomography, and RVP measured using an ophthalmodynamometer were obtained before intravitreal injection of ranibizumab (IVR) and 1 month later.

Results

Comparison of the BRVOs and CRVOs showed that VA was significantly improved by a single injection in BRVOs (P < 0.0001; P = 0.1087 for CRVOs), but CRT and RVP were significantly decreased without significant difference in IOP after the treatment in both groups (P < 0.0001).

Conclusion

The anti-VEGF treatment resulted in a significant decrease in the RVP, but the RVP remained significantly higher than the IOP. An increased RVP plays a decisive role in the formation of macula edema, and reducing it is desirable.