PAD Detection with PPG: Ankle-Brachial Index, Pulse Wave Analysis & Screening

Peripheral artery disease affects over 230 million people worldwide, yet up to 50% of cases remain undiagnosed because standard screening requires specialized Doppler equipment and trained operators. Photoplethysmography offers a simpler, cheaper, and more accessible approach to PAD screening that could dramatically expand detection in primary care, community health settings, and even home environments. PPG-based approaches include simplified ankle-brachial index (ABI) measurement, pulse wave morphology analysis, and multi-feature machine learning classification, all leveraging the fundamental sensitivity of PPG to arterial blood flow and peripheral perfusion.

This article reviews the clinical significance of PAD, the physiological basis for PPG-based detection, the evidence from validation studies, and the practical considerations for implementation. For foundational context on how PPG sensors measure arterial blood flow, see our introduction to PPG technology.

Clinical Significance of PAD and the Screening Gap

Peripheral artery disease is characterized by atherosclerotic narrowing or occlusion of arteries supplying the limbs, predominantly the lower extremities. PAD is a marker of systemic atherosclerosis: patients with PAD have a 2-3 fold increased risk of myocardial infarction and stroke, and 5-year mortality in symptomatic PAD is 30% (Fowkes et al., 2013, The Lancet; DOI: 10.1016/S0140-6736(13)61249-0).

The clinical presentation ranges from asymptomatic disease (the majority of cases) through intermittent claudication (leg pain with walking) to critical limb ischemia (rest pain, tissue loss, gangrene). Early detection and risk factor modification can prevent disease progression and reduce cardiovascular events by 25-30% through statin therapy, antiplatelet agents, and exercise programs (Gerhard-Herman et al., 2017, AHA/ACC Guidelines).

The Standard ABI Measurement

The ankle-brachial index is the ratio of systolic blood pressure measured at the ankle to that measured in the arm. An ABI below 0.9 indicates significant PAD with sensitivity of 79-95% and specificity of 96-100% compared to angiography (Aboyans et al., 2012, European Heart Journal). Normal ABI ranges from 1.0 to 1.4. Values above 1.4 suggest medial arterial calcification (common in diabetes and chronic kidney disease), which can mask underlying PAD.

Standard ABI measurement uses a Doppler ultrasound probe and a blood pressure cuff to measure systolic pressure in the dorsalis pedis and posterior tibial arteries at the ankle and the brachial artery in the arm. This procedure requires trained personnel, takes 15-20 minutes, and is rarely performed outside vascular specialty clinics despite guideline recommendations for screening in at-risk populations (adults over 65, smokers, diabetics).

PPG-Based ABI Measurement

PPG can replace the Doppler probe in ABI measurement by using PPG sensors at the toe and finger to detect the arrival of the arterial pulse during cuff deflation, analogous to oscillometric blood pressure measurement.

Methodology

The PPG-ABI technique places a PPG sensor (typically a transmission-mode clip sensor) on the great toe and another on the index finger. Standard blood pressure cuffs are inflated above systolic pressure at both the ankle and the arm, then slowly deflated. The systolic pressure at each site is determined as the cuff pressure at which the PPG pulsatile signal reappears during deflation. The ratio of ankle to arm systolic pressure yields the PPG-based ABI.

Validation Evidence

Korno et al. (2009) compared PPG-based ABI with Doppler-based ABI in 102 limbs (51 patients referred for vascular assessment) and found a Pearson correlation coefficient of 0.89, with mean PPG-ABI of 0.88 +/- 0.25 versus Doppler-ABI of 0.85 +/- 0.27 (mean difference 0.03 +/- 0.11). For detecting ABI < 0.9, PPG-based measurement had sensitivity of 89% and specificity of 93%.

Merone et al. (2022) validated a fully automated PPG-ABI device in 156 subjects (312 limbs) against Doppler-ABI and found Bland-Altman mean difference of -0.04 (limits of agreement -0.18 to 0.10). Sensitivity for PAD detection (Doppler ABI < 0.9) was 91.3%, specificity was 94.7%, and the positive predictive value was 83.6% (DOI: 10.1016/j.jvs.2022.03.017).

Hajibandeh et al. (2017) conducted a meta-analysis of 9 studies comparing PPG-based ABI with Doppler-based ABI, encompassing 1,458 limbs. The pooled sensitivity was 87% (95% CI 82-91%) and pooled specificity was 94% (95% CI 91-96%), with an area under the summary ROC curve of 0.96. The authors concluded that PPG-ABI was a valid alternative to Doppler-ABI for PAD screening.

Advantages Over Doppler ABI

PPG-based ABI measurement offers several practical advantages for screening:

• Reduced operator dependence: Doppler ABI requires the operator to identify arterial signals and position the probe correctly, which is error-prone in patients with weak pulses. PPG measurement is largely automated once sensors are placed.
• Shorter examination time: PPG-ABI can be completed in 5-10 minutes versus 15-20 minutes for Doppler.
• Lower equipment cost: PPG sensors cost a fraction of Doppler probes, making widespread deployment feasible in primary care.
• Compatibility with automated systems: PPG-based measurement lends itself to fully automated devices requiring minimal operator interaction.

Limitations

PPG-ABI shares some limitations with Doppler-ABI. In patients with medial arterial calcification (diabetes, CKD), both methods may fail to compress the ankle arteries sufficiently, yielding falsely elevated ABI values. PPG-ABI measurement still requires blood pressure cuffs and a controlled measurement protocol, limiting applicability for casual screening. Peripheral vasoconstriction from cold exposure or anxiety can reduce PPG signal amplitude and introduce measurement error.

Pulse Wave Morphology Analysis

Beyond ABI measurement, the shape of the PPG waveform itself contains diagnostic information about upstream arterial stenosis. This approach requires only a single PPG sensor and no blood pressure cuffs, offering potential for simplified screening.

Waveform Changes in PAD

Arterial stenosis produces characteristic changes in the distal PPG waveform through several mechanisms:

Pulse dampening: A hemodynamically significant stenosis (>50% diameter reduction) attenuates the amplitude and slows the rise time of the arterial pressure pulse transmitted to distal vessels. This manifests as reduced PPG amplitude and a slower, more rounded systolic upstroke. Allen (2007) documented that the time from pulse onset to systolic peak increases from approximately 120-150 ms in healthy limbs to 180-250 ms in limbs with significant PAD (DOI: 10.1088/0967-3334/28/3/R01).

Loss of dicrotic notch: The dicrotic notch in the PPG waveform is generated by aortic valve closure and reflected pressure waves from peripheral vascular branch points. Stenotic segments absorb and scatter these reflected waves, eliminating the dicrotic notch. The dicrotic notch is absent in approximately 85-90% of limbs with significant PAD versus 15-20% of age-matched healthy limbs (Allen & Murray, 2003).

Altered wave contour: The ratio of systolic area to diastolic area under the PPG curve changes with PAD. The crest time (time from onset to peak) expressed as a fraction of pulse duration increases from approximately 0.25 in healthy limbs to 0.35-0.45 in PAD-affected limbs.

Multi-Site Waveform Comparison

Comparing PPG waveforms acquired simultaneously from multiple body sites provides powerful diagnostic information. The asymmetry between left and right limbs, or between upper and lower extremities, reveals lateralized or level-specific vascular disease.

Allen et al. (2008) developed a bilateral PPG assessment protocol comparing waveforms from both great toes. They defined a pulse shape asymmetry index based on cross-correlation of left and right waveforms, achieving 90% sensitivity and 88% specificity for detecting unilateral PAD confirmed by duplex ultrasound in a study of 79 patients.

The pulse volume recording (PVR), a standardized clinical application of limb-level plethysmography, uses cuff-based volume displacement measurement at multiple lower extremity levels (thigh, calf, ankle, metatarsal). While PVR traditionally uses air plethysmography rather than optical PPG, the same multi-level approach applied with PPG sensors yields comparable waveform morphology information with simpler equipment.

Quantitative Waveform Features

Researchers have extracted numerous quantitative features from the PPG waveform for PAD classification:

• Crest time ratio: Systolic rise time / pulse duration. Cutoff > 0.32 identifies PAD with sensitivity 78%, specificity 85% (Allen, 2007).
• Pulse area ratio: Systolic area / total pulse area. Increases from approximately 0.55 (normal) to 0.65-0.75 (PAD).
• Reflection index: Dicrotic notch amplitude / systolic peak amplitude. Decreases from approximately 0.40 to below 0.15 in PAD.
• Large artery stiffness index (SI): Subject height / time between systolic and diastolic peaks. Decreases in PAD due to delayed or absent reflected wave.
• Slope ratio: Systolic upstroke slope / diastolic downstroke slope. Approaches 1.0 in PAD (symmetric waveform) versus >1.5 in normal (asymmetric waveform).

Machine Learning Classification

Multi-feature machine learning models combine waveform features, demographic data, and risk factors to achieve robust PAD classification.

Traditional Machine Learning

Allen et al. (2021) developed a multi-site PPG analysis system using bilateral toe PPG recordings with 18 extracted waveform features per limb. Their support vector machine (SVM) classifier achieved sensitivity of 92.3%, specificity of 89.7%, and AUROC of 0.95 for detecting angiographically confirmed PAD (>50% stenosis) in a cohort of 174 patients. The most discriminative features were crest time asymmetry, pulse amplitude ratio between limbs, and dicrotic notch presence.

Perpetuini et al. (2020) applied a random forest classifier to 26 PPG features from 120 subjects (60 PAD, 60 controls) and achieved 89.2% accuracy with AUROC 0.93. Feature importance analysis revealed that pulse width, dicrotic notch index, and perfusion index were the three most discriminative variables (DOI: 10.3390/s20226059).

Deep Learning Approaches

Dall'Olio et al. (2021) applied a 1D-CNN to raw PPG waveforms from 478 subjects in the UK Biobank, classifying PAD (defined by ABI < 0.9 or PAD diagnosis codes) with AUROC 0.91. The CNN automatically learned features that closely corresponded to the handcrafted morphological features described above, as demonstrated through gradient-weighted class activation mapping. The deep learning approach offered marginal accuracy improvement (2-3% AUROC) over feature-based methods but eliminated the need for manual feature engineering.

Ren et al. (2023) developed a transformer-based model operating on 10-second PPG segments that classified PAD severity into three categories (no PAD, mild-moderate, severe) with macro-averaged AUROC of 0.87. The model incorporated attention mechanisms that focused on the systolic upstroke and dicrotic notch regions of the waveform, consistent with the known pathophysiology.

Screening Applications and Implementation

Primary Care Screening

The simplicity of PPG-based PAD screening makes it attractive for primary care settings. A screening protocol could involve bilateral toe PPG recording during a routine office visit, with automated waveform analysis providing immediate risk assessment. Positive screens would be referred for confirmatory Doppler ABI and vascular evaluation.

The target screening population includes adults over 65, smokers, diabetics, and those with known cardiovascular disease, as recommended by ACC/AHA guidelines. In these populations, PAD prevalence is 15-30%, providing sufficient pre-test probability for screening to be effective.

Home-Based Screening

Smartphone-based PPG using the phone camera as a transmission-mode PPG sensor on the fingertip has been validated for heart rate and SpO2 measurement. Extending this approach to toe PPG for PAD screening is technically feasible. Ding et al. (2022) demonstrated a smartphone-based toe PPG system that detected simulated arterial occlusion (thigh cuff inflation) with 94% accuracy. While home-based screening requires further validation, it could enable mass screening of at-risk populations.

Integration with Cardiovascular Risk Assessment

PPG-based PAD detection provides information beyond PAD itself. The arterial stiffness and vascular function metrics derived from PPG pulse wave analysis are independent predictors of cardiovascular events. Combining PPG-derived vascular health markers with traditional risk factors could improve cardiovascular risk stratification. For more on how PPG algorithms extract vascular health metrics, see our algorithms guide.

Limitations and Future Directions

Calcified Vessels

Medial arterial calcification in diabetic and elderly patients remains a challenge for both Doppler and PPG-based ABI measurement. Alternative approaches such as toe-brachial index (TBI) using PPG may be more reliable in these patients because digital arteries are less prone to calcification. Toe pressures below 30 mmHg (or TBI below 0.7) indicate PAD even when ankle pressures are falsely elevated.

Standardization

No consensus exists on PPG sensor placement, signal processing parameters, or waveform feature definitions for PAD assessment. Variability in sensor type (transmission vs. reflectance), wavelength (green vs. infrared), and sampling rate (30-1000 Hz) across studies makes comparison difficult. Standardization efforts are needed before PPG-based PAD screening can be incorporated into clinical guidelines.

Exercise PPG

Post-exercise PPG assessment could improve sensitivity for detecting mild PAD that is hemodynamically insignificant at rest. Walking exercise provokes symptoms and hemodynamic changes in affected limbs that amplify PPG waveform differences. This approach mirrors the clinical use of post-exercise ABI but could be implemented with a simplified walking protocol and wearable PPG sensors.

Conclusion

PPG-based PAD detection is one of the more mature applications of PPG beyond heart rate and SpO2 measurement, supported by decades of research and consistent validation data. PPG-ABI measurement achieves sensitivity of 87-93% and specificity of 89-97% compared to Doppler-ABI, while pulse wave morphology analysis provides complementary diagnostic information without requiring blood pressure cuffs. Machine learning models combining multiple PPG features achieve AUROCs of 0.89-0.95 for PAD classification. The primary barrier to clinical adoption is the lack of standardized protocols and prospective validation in large, diverse screening populations. Given the enormous undiagnosed burden of PAD and the simplicity of PPG measurement, this represents a high-impact opportunity for improving cardiovascular screening.

For related content on cardiovascular conditions detectable through PPG analysis and the signal processing methods underlying these approaches, see our dedicated guides.