PPG Waveform Morphology Features: Physiological Meaning and Clinical Applications

The shape of a photoplethysmographic pulse waveform is not merely a stylized heartbeat trace. Every contour, slope, and inflection point encodes specific information about the cardiovascular system that generated it. The systolic upstroke reflects left ventricular ejection dynamics. The dicrotic notch marks aortic valve closure. The diastolic decay reveals arterial compliance and peripheral resistance. The reflected wave component indicates arterial stiffness and wave reflection magnitude. Extracting and quantifying these morphological features transforms the PPG signal from a simple heart rate source into a window on cardiovascular health.

This article provides a comprehensive technical guide to PPG waveform morphology features, their physiological origins, extraction methods, and clinical applications. For foundational understanding of how PPG signals are generated, see our introduction to photoplethysmography. For information on how the AC and DC components relate to these features, see our guide on the AC/DC ratio in pulse oximetry.

Anatomy of the PPG Pulse Waveform

A single PPG pulse waveform, representing one cardiac cycle, contains several identifiable landmarks and regions. Understanding each is essential before extracting quantitative features.

The Systolic Phase

The systolic phase begins at the foot of the waveform (the trough between consecutive pulses) and extends to the systolic peak. This ascending limb reflects the arrival of the cardiac pulse wave at the measurement site.

The systolic upstroke is the rising portion from the foot to the peak. Its slope (dP/dt, or more precisely dI/dt where I is PPG intensity) is related to the rate of increase in local arterial blood volume, which depends on left ventricular contractility, proximal arterial compliance, and the propagation characteristics of the pulse wave. A steeper upstroke suggests higher pulse pressure and faster pulse wave arrival. Millasseau et al. (2006) showed that the maximum slope of the systolic upstroke correlated with aortic pulse wave velocity (r = 0.58, p < 0.001, n = 68), linking this simple morphological feature to central arterial stiffness (DOI: 10.1097/01.hjh.0000239286.02181.e0).

The systolic peak is the maximum amplitude point of the pulse waveform. Its amplitude relative to the DC baseline defines the Perfusion Index. In subjects with stiff arteries, the systolic peak may be augmented by an early-returning reflected wave, producing a higher and sometimes broader peak than would be generated by the forward-traveling wave alone.

The Dicrotic Notch

The dicrotic notch is a small but physiologically significant deflection on the descending limb of the waveform. It corresponds to the momentary reversal of aortic flow and subsequent aortic valve closure at the end of ventricular systole. In central arterial pressure waveforms (measured invasively in the aorta), the dicrotic notch is sharp and well-defined. In peripheral PPG waveforms, it is attenuated by pulse wave damping and distortion during transmission through the arterial tree.

The visibility and prominence of the dicrotic notch vary with several factors:

• Age: The dicrotic notch is most prominent in young adults (ages 20-35) with compliant arteries and diminishes or disappears entirely in older adults. Millasseau et al. (2002) documented a systematic decrease in dicrotic notch prominence with age, with the notch becoming undetectable in many subjects over age 60 (DOI: 10.1161/01.HYP.0000016936.68948.ce).
• Arterial stiffness: Independent of age, increased arterial stiffness reduces dicrotic notch prominence by increasing pulse wave velocity, which causes the reflected wave to arrive earlier and merge with the systolic wave.
• Heart rate: Higher heart rates compress the diastolic interval, reducing the time window in which the notch is visible.
• Measurement site: The dicrotic notch is more prominent in proximal arteries (brachial, radial) than distal sites (fingertip), and is often barely detectable in wrist PPG.

The Diastolic Phase

The diastolic phase extends from the dicrotic notch to the foot of the next pulse. The diastolic decay curve reflects the runoff of blood from the arterial system through the peripheral resistance vessels. In some subjects, particularly younger ones, a distinct diastolic peak (also called the reflected wave or diastolic wave) is visible as a secondary peak or inflection on the diastolic decay.

The diastolic peak arises from pulse wave reflection at sites of impedance mismatch in the arterial tree, primarily at arterial bifurcations and small muscular arteries. The timing and amplitude of this reflected wave carry important information about arterial stiffness and wave reflection magnitude.

Key Morphological Features and Indices

Augmentation Index (AIx)

The Augmentation Index quantifies the contribution of the reflected wave to the systolic pressure:

AIx = (P2 - P1) / PP x 100

Where P2 is the amplitude at the inflection point (where the reflected wave arrives), P1 is the amplitude of the early systolic peak (before the reflected wave contribution), and PP is the total pulse pressure (systolic peak - diastolic foot).

In young adults with compliant arteries, the reflected wave arrives during diastole (after the systolic peak), so AIx is negative or near zero. In older adults with stiffer arteries, faster pulse wave velocity causes the reflected wave to arrive during late systole, augmenting the systolic peak and producing a positive AIx.

Wilkinson et al. (1998) established that radial artery AIx derived from tonometry correlated strongly with aortic AIx (r = 0.91) and increased linearly with age at approximately 0.5% per year between ages 20 and 80. While PPG-derived AIx is not identical to tonometric AIx due to differences in measurement modality, the correlation between PPG AIx and age has been validated in multiple studies, with reported correlation coefficients of 0.55-0.75.

Reflection Index (RI)

The Reflection Index quantifies the relative amplitude of the diastolic (reflected) wave compared to the systolic peak:

RI = P_diastolic / P_systolic x 100

Where P_diastolic is the amplitude of the diastolic peak and P_systolic is the amplitude of the systolic peak, both measured from the waveform foot.

RI reflects the magnitude of pulse wave reflection from the peripheral vasculature. Higher RI indicates stronger wave reflection, which is associated with increased peripheral vascular resistance and arterial stiffness. Millasseau et al. (2006) found that RI measured from the digital volume pulse (DVP, which is essentially fingertip PPG) correlated with brachial-ankle pulse wave velocity (r = 0.48, p < 0.001) and with age (r = -0.37, p < 0.01). The negative correlation with age reflects the fact that while reflection sites become more proximal with aging, the magnitude of reflection at any given site tends to decrease as arteries stiffen.

Stiffness Index (SI)

The Stiffness Index provides an estimate of aortic pulse wave velocity from the timing of the reflected wave in the PPG waveform:

SI = body height / delta-T

Where delta-T is the time delay between the systolic peak and the diastolic peak (reflected wave). The rationale is that delta-T represents the transit time of the pulse wave from the heart to the major reflection site and back, and body height serves as a proxy for this path length.

SI was introduced by Millasseau et al. (2002) using digital volume pulse analysis and has been validated against carotid-femoral pulse wave velocity (cfPWV), the gold standard measure of arterial stiffness. They reported a correlation of r = 0.65 (p < 0.001, n = 87) between SI and cfPWV. SI increases with age and is elevated in conditions associated with arterial stiffening, including hypertension, diabetes, and chronic kidney disease. Typical values range from 5-7 m/s in healthy young adults to 10-15 m/s in elderly or hypertensive individuals. For more on pulse wave velocity measurement, see our dedicated article on PWV as an arterial stiffness marker.

Crest Time and Pulse Width

Crest time (CT) is the time from the waveform foot to the systolic peak. It reflects the rate of rise of the arterial pressure pulse and is influenced by left ventricular ejection dynamics and proximal arterial compliance. Shorter crest times are associated with faster pulse wave propagation and stiffer arteries.

Pulse width at various amplitude fractions (typically 10%, 25%, 50%, and 75% of peak amplitude) characterizes the overall shape of the waveform. Wider pulses at lower amplitude fractions suggest slower diastolic runoff and lower peripheral resistance. Wang et al. (2013) used pulse width features in machine learning models for blood pressure estimation, reporting correlations of r = 0.54-0.67 with systolic blood pressure across a dataset of 1,276 subjects (DOI: 10.1109/TBME.2012.2213727).

Area Ratios

The ratio of the systolic area (from foot to dicrotic notch) to the diastolic area (from dicrotic notch to next foot) reflects the balance between cardiac ejection and diastolic runoff. This systolic-to-diastolic area ratio is related to total peripheral resistance and arterial compliance:

Area Ratio = A_systolic / A_diastolic

A higher area ratio suggests a waveform dominated by the systolic ejection phase, which can indicate reduced arterial compliance. Conversely, a lower area ratio (relatively more area under the diastolic curve) suggests better arterial compliance and lower vascular resistance. Elgendi (2012) incorporated area ratio features into PPG classification algorithms and found them among the most discriminative features for differentiating normal from abnormal cardiovascular states (DOI: 10.1371/journal.pone.0076585).

Second Derivative Analysis (SDPPG)

The second derivative of the PPG waveform, known as the acceleration plethysmogram (APG) or SDPPG, is one of the most extensively studied tools for extracting morphological information from PPG signals.

The Five Characteristic Waves

The SDPPG of a typical PPG pulse contains five identifiable waves, labeled a, b, c, d, and e by Takazawa et al. (1998) in their seminal paper (DOI: 10.1038/sj.jhh.1000818):

• Wave a: The initial positive wave, corresponding to the early systolic acceleration of blood flow. It is the largest positive deflection and serves as the reference for ratio calculations.
• Wave b: The first negative wave, following a. It reflects the deceleration of the late systolic flow and the reduction in arterial distension rate. The b/a ratio increases with arterial stiffness because stiffer arteries transmit the pressure wave faster, creating a sharper transition from acceleration to deceleration.
• Wave c: The second positive wave, appearing in mid-systole. It corresponds to the re-acceleration caused by the reflected wave arrival (in subjects with moderate arterial stiffness) or the continued systolic flow pattern.
• Wave d: The second negative wave, corresponding to the dicrotic notch region. Its amplitude reflects the prominence of the dicrotic notch and therefore arterial compliance.
• Wave e: The final positive wave in the diastolic phase, corresponding to the diastolic reflected wave.

SDPPG Indices and Vascular Aging

Takazawa et al. (1998) defined an aging index from the SDPPG wave amplitudes:

Aging Index (AGI) = (b - c - d - e) / a

Where each letter represents the amplitude of the corresponding wave (positive waves have positive values, negative waves have negative values). This index showed a correlation of r = 0.80 with chronological age in a study of 600 subjects aged 20-79 years, making it one of the strongest age correlations reported for any non-invasive vascular measurement.

Additional SDPPG ratios have been validated:

• b/a ratio: Increases with age and arterial stiffness. Correlates with brachial-ankle PWV (r = 0.53-0.66 across studies). Values range from approximately -0.3 in young adults to -0.8 in elderly subjects (note: b is typically negative, so b/a is negative, and more negative values indicate greater stiffness).
• c/a ratio: Decreases with age. Reflects the timing and magnitude of wave reflection.
• d/a ratio: Decreases with age. Related to dicrotic notch prominence and arterial compliance.
• e/a ratio: Variable with age. Related to the diastolic wave component.

Imanaga et al. (1998) validated SDPPG indices against invasive aortic pressure waveform analysis and found that the SDPPG-derived b/a ratio correlated with the augmentation index measured from the aortic waveform (r = 0.67, p < 0.001, n = 55). For clinical context on how arterial stiffness relates to cardiovascular conditions, see our conditions page.

Feature Extraction Methods

Time-Domain Feature Extraction

Time-domain feature extraction involves identifying specific fiducial points (landmarks) on the PPG waveform and computing temporal and amplitude relationships between them.

The critical first step is accurate identification of the systolic foot, systolic peak, dicrotic notch, and diastolic peak. Peak detection algorithms range from simple threshold crossings to sophisticated template-matching methods. Our peak detection algorithms guide covers these techniques in detail.

The dicrotic notch is particularly challenging to detect automatically because it varies widely in prominence and may be absent in waveforms from older subjects or those with stiff arteries. Common detection approaches include:

1. First derivative zero-crossing: The dicrotic notch appears as a local minimum in the first derivative of the PPG waveform, detectable as a zero-crossing from negative to positive in the descending limb.
2. Second derivative analysis: The dicrotic notch corresponds to the d-wave in the SDPPG, allowing its timing to be identified from SDPPG landmarks.
3. Gaussian decomposition: Modeling the PPG pulse as a sum of Gaussian functions (typically 3-5), where one Gaussian corresponds to the dicrotic notch region. Liu et al. (2013) demonstrated that Gaussian decomposition into four components provided robust feature extraction across different subject populations (DOI: 10.1016/j.bspc.2013.05.007).

Frequency-Domain Analysis

Fourier decomposition of the PPG waveform into harmonic components provides an alternative feature set. The relative amplitudes and phases of the first 5-10 harmonics capture the waveform shape in a compact representation. The first harmonic amplitude reflects the fundamental pulse amplitude, while higher harmonics encode waveform sharpness and complexity.

Wang et al. (2015) used the harmonic-to-fundamental ratio (amplitude of higher harmonics relative to the first harmonic) as a measure of waveform complexity and showed it correlated with arterial stiffness indices. This frequency-domain approach has the advantage of being less dependent on precise fiducial point detection, which can be error-prone in noisy or low-quality signals.

Machine Learning Feature Extraction

Modern approaches increasingly use machine learning to extract features from PPG waveforms without requiring explicit fiducial point identification. Convolutional neural networks can learn relevant waveform features directly from raw or minimally preprocessed PPG segments.

Slapnicar et al. (2019) applied deep learning with spectro-temporal ResNet architecture to PPG waveforms for blood pressure estimation, achieving mean absolute errors of 9.43 mmHg for systolic and 6.88 mmHg for diastolic blood pressure on a dataset of 510 subjects (DOI: 10.3390/s19153420). The network learned features that corresponded to known morphological indices but also captured subtle patterns not captured by traditional handcrafted features. For more information on how machine learning is applied to PPG signals, see our algorithms overview.

Clinical Applications of Morphology Analysis

Vascular Age Estimation

PPG morphology-based vascular age estimation is one of the most mature clinical applications. By comparing a subject's PPG morphological indices (particularly SDPPG aging index, augmentation index, and stiffness index) against age-stratified reference populations, a "vascular age" can be estimated that reflects the functional state of the arterial system rather than chronological age.

A vascular age significantly exceeding chronological age suggests premature arterial aging, which is a risk factor for cardiovascular events. Conversely, a vascular age below chronological age may indicate favorable vascular health. Bortolotto et al. (2000) showed that PPG-derived vascular age indices correlated with established cardiovascular risk factors including hypertension, diabetes, hyperlipidemia, and smoking status (DOI: 10.1038/sj.jhh.1001045).

Atrial Fibrillation Detection

While atrial fibrillation detection from PPG primarily relies on heart rate irregularity (absence of the regular sinus rhythm pattern), morphological features provide additional discriminative power. During AF, PPG pulses show beat-to-beat variability in amplitude, shape, and timing that is distinct from the regular morphology variation seen in normal sinus rhythm with respiratory sinus arrhythmia.

Kwon et al. (2019) combined PPG morphology features (pulse amplitude variability, waveform complexity measures) with rhythm irregularity metrics and achieved 98.2% sensitivity and 98.1% specificity for AF detection in a study of 508 subjects. The morphological features contributed an incremental improvement of 2-3% in accuracy over rhythm analysis alone (DOI: 10.1038/s41746-019-0130-7). For more on how AF detection works from PPG, see our AF detection algorithms page.

Sleep Staging and Autonomic Assessment

PPG morphology changes during different sleep stages, reflecting the autonomic nervous system shifts that accompany sleep architecture transitions. During deep slow-wave sleep, sympathetic withdrawal leads to vasodilation, increasing pulse amplitude and shifting morphological indices. During REM sleep, sympathetic surges can cause transient vasoconstriction and morphology changes.

Measuring these morphological shifts over a night of sleep provides a non-invasive window into autonomic balance that complements heart rate variability analysis. Wrist-worn wearables that capture continuous PPG during sleep are positioned to exploit these morphological features for improved sleep staging and autonomic health assessment.

Challenges and Limitations

Signal Quality Dependence

Morphology analysis requires substantially higher signal quality than simple heart rate extraction. Heart rate can be estimated from heavily corrupted signals using frequency-domain methods, but morphological features require accurate representation of the waveform shape, including small features like the dicrotic notch. Signal quality assessment is therefore essential before morphology analysis. For techniques to handle noisy PPG signals, see our motion artifact removal guide.

Measurement Site Variability

PPG waveform morphology varies significantly across body sites due to pulse wave transmission effects. The waveform at the fingertip differs from the earlobe, which differs from the wrist. Features extracted at one site cannot be directly compared with reference values established at another site without site-specific calibration or normalization. This is particularly relevant for wrist-based wearables, where the PPG waveform is typically more damped and less feature-rich than fingertip recordings.

Individual Variability

Even at a fixed measurement site, PPG morphology varies substantially between individuals due to differences in arterial anatomy, body composition, skin properties, and cardiovascular physiology. Population-level correlations between morphological features and clinical parameters (e.g., blood pressure, arterial stiffness) may not translate to accurate individual-level estimation. Establishing personalized baselines and tracking within-individual trends is generally more robust than applying population-derived thresholds.

Summary

PPG waveform morphology analysis extends the information extractable from photoplethysmography far beyond heart rate and SpO2. By quantifying features of the pulse waveform shape, including systolic upstroke characteristics, dicrotic notch prominence, reflected wave timing and amplitude, and second derivative wave patterns, researchers and clinicians can assess arterial stiffness, vascular age, autonomic function, and potentially estimate blood pressure non-invasively. While challenges remain in signal quality requirements, measurement site standardization, and individual variability, the growing deployment of high-quality PPG sensors in clinical and consumer devices is accelerating the translation of morphology analysis from research to practice.

Frequently Asked Questions

What is the dicrotic notch in a PPG waveform?

The dicrotic notch is a small downward deflection that appears on the descending (diastolic) portion of the PPG pulse waveform. It corresponds to the momentary retrograde flow and subsequent aortic valve closure at the end of ventricular systole. In PPG, the dicrotic notch is attenuated compared to central arterial waveforms because the pulse wave undergoes distortion as it travels to the periphery. The prominence of the dicrotic notch decreases with age and arterial stiffness: young adults with compliant arteries show a distinct notch, while older adults with stiff arteries often show a smooth, notch-free descending limb.

What does the second derivative of a PPG signal (SDPPG) reveal?

The second derivative of the PPG signal (SDPPG or acceleration plethysmogram) amplifies inflection points in the original waveform, making subtle morphological features more visible and quantifiable. The SDPPG typically contains five characteristic waves labeled a through e, whose relative amplitudes reflect arterial stiffness and vascular aging. The b/a ratio (ratio of the second to first wave amplitude) increases with arterial stiffness, while the d/a ratio decreases. Takazawa et al. (1998) demonstrated that the aging index (b-c-d-e)/a derived from SDPPG correlates strongly with vascular age (r = 0.80) and has been validated as a non-invasive arterial stiffness marker.

Can PPG waveform features estimate blood pressure?

PPG waveform morphology features have been extensively studied for cuffless blood pressure estimation, but with limited clinical success so far. Features like pulse width, diastolic peak timing, reflection index, and area ratios correlate with blood pressure in population studies, but the correlations (typically r = 0.3-0.6) are insufficient for individual-level blood pressure tracking. The primary challenge is that PPG morphology is influenced by multiple physiological variables simultaneously (arterial stiffness, vascular tone, stroke volume, heart rate), making it difficult to isolate the blood pressure contribution. Current research focuses on combining morphology features with machine learning and pulse transit time for improved accuracy.

How does age affect PPG waveform shape?

Aging produces characteristic changes in PPG waveform morphology driven by progressive arterial stiffening. In young adults (ages 20-30), PPG waveforms show a sharp systolic peak, a clear dicrotic notch, and a prominent diastolic peak. The augmentation index is low and the reflection index is moderate. With aging, arterial pulse wave velocity increases, causing the reflected wave to return earlier and merge with the systolic peak. By age 60-70, the dicrotic notch often disappears, the waveform becomes more triangular, and augmentation index increases. These changes are well-documented and form the basis of PPG-based vascular age estimation.

References

• The systolic upstroke* is the rising portion from the foot to the peak. Its slope (dP/dt, or more precisely dI/dt where I is PPG intensity) is related to the rate of increase in local arterial blood volume, which depends on left ventricular contractility, proximal arterial compliance, and the propagation characteristics of the pulse wave. A steeper upstroke suggests higher pulse pressure and faster pulse wave arrival. Millasseau et al. (2006) showed that the maximum slope of the systolic upstroke correlated with aortic pulse wave velocity (r = 0.58, p < 0.001, n = 68), linking this simple morphological feature to central arterial stiffness (DOI: 10.1097/01.hjh.0000239286.02181.e0).
• Age: The dicrotic notch is most prominent in young adults (ages 20-35) with compliant arteries and diminishes or disappears entirely in older adults. Millasseau et al. (2002) documented a systematic decrease in dicrotic notch prominence with age, with the notch becoming undetectable in many subjects over age 60 (DOI: 10.1161/01.HYP.0000016936.68948.ce).
• Pulse width* at various amplitude fractions (typically 10%, 25%, 50%, and 75% of peak amplitude) characterizes the overall shape of the waveform. Wider pulses at lower amplitude fractions suggest slower diastolic runoff and lower peripheral resistance. Wang et al. (2013) used pulse width features in machine learning models for blood pressure estimation, reporting correlations of r = 0.54-0.67 with systolic blood pressure across a dataset of 1,276 subjects (DOI: 10.1109/TBME.2012.2213727).
• A higher area ratio suggests a waveform dominated by the systolic ejection phase, which can indicate reduced arterial compliance. Conversely, a lower area ratio (relatively more area under the diastolic curve) suggests better arterial compliance and lower vascular resistance. Elgendi (2012) incorporated area ratio features into PPG classification algorithms and found them among the most discriminative features for differentiating normal from abnormal cardiovascular states (DOI: 10.1371/journal.pone.0076585).
• The SDPPG of a typical PPG pulse contains five identifiable waves, labeled a, b, c, d, and e by Takazawa et al. (1998) in their seminal paper (DOI: 10.1038/sj.jhh.1000818):
• 3. Gaussian decomposition: Modeling the PPG pulse as a sum of Gaussian functions (typically 3-5), where one Gaussian corresponds to the dicrotic notch region. Liu et al. (2013) demonstrated that Gaussian decomposition into four components provided robust feature extraction across different subject populations (DOI: 10.1016/j.bspc.2013.05.007).
• Slapnicar et al. (2019) applied deep learning with spectro-temporal ResNet architecture to PPG waveforms for blood pressure estimation, achieving mean absolute errors of 9.43 mmHg for systolic and 6.88 mmHg for diastolic blood pressure on a dataset of 510 subjects (DOI: 10.3390/s19153420). The network learned features that corresponded to known morphological indices but also captured subtle patterns not captured by traditional handcrafted features. For more information on how machine learning is applied to PPG signals, see our algorithms overview.
• A vascular age significantly exceeding chronological age suggests premature arterial aging, which is a risk factor for cardiovascular events. Conversely, a vascular age below chronological age may indicate favorable vascular health. Bortolotto et al. (2000) showed that PPG-derived vascular age indices correlated with established cardiovascular risk factors including hypertension, diabetes, hyperlipidemia, and smoking status (DOI: 10.1038/sj.jhh.1001045).
• Kwon et al. (2019) combined PPG morphology features (pulse amplitude variability, waveform complexity measures) with rhythm irregularity metrics and achieved 98.2% sensitivity and 98.1% specificity for AF detection in a study of 508 subjects. The morphological features contributed an incremental improvement of 2-3% in accuracy over rhythm analysis alone (DOI: 10.1038/s41746-019-0130-7). For more on how AF detection works from PPG, see our AF detection algorithms page.