Eigenpulse: Eliminating Demographic Bias in Pulse Oximetry and Remote PPG from First Principles

Towards the world's first non-contact co-oximeter

In 2024, my father was hospitalized and I watched overworked staff barely checking his vitals. I built a camera-based system using remote photoplethysmography (rPPG)1 to monitor patients without contact2(Fig-1). It worked in labs, but when deployed in nursing homes, motion caused significant failures. Investigating why exposed a deeper problem.

The field reports ~14 beats-per-minute (BPM) error on dark skin versus ~5 BPM on light3. The same bias exists in contact pulse oximetry, documented for decades, yet unchanged4-6. During COVID-19, Black patients experienced three times the rate of undetected hypoxemia6, contributing to a 41% increase in mortality7(Fig-2). The response has been ever larger models, yet cross-dataset accuracy remains 9-10 BPM (Fig-3).The FDA's 2025 draft guidance8 scales testing to 150 diverse participants, yet has no theory for what manufacturers should fix. Experts state the device physics must be corrected9.

The field blames lower signal-to-noise-ratio (SNR) due to melanin absorption3. I asked a different question: what if the signal model itself is incomplete? Each heartbeat changes blood volume, creating two distinct optical effects. The first is wavelength-dependent absorption3, which current models capture. The second is the physical expansion of tissue that modulates scattered light at cardiac frequency across all wavelengths, which they do not (Fig-4). They also overlook a multiplicative noise: vascular interference coupled to the signal itself10.

All roads led back to incomplete physics. Correcting it opened five discoveries, three innovations, and two journal papers under peer review11-13(Fig-5). Improved physics is what will separate every patient seen from some systematically missed.

Current rPPG methods (CHROM18, POS19) project RGB onto fixed directions to isolate the cardiac pulse. My geometric model shows that melanin rotates the isochromatic contaminant into the measurement plane, where it mixes with the cardiac signal. Across 720 regions-of-interest (ROI) in 3 cohorts and 2 datasets, theoretical curves with zero free parameters match measurement at Pearson r = 0.997 (POS) and r = 0.973 (CHROM). The black lines in Fig-6 are not fits. Demographic bias is not stochastic; it is structural.

The failure compounds temporally. Current methods adaptively cancel the dominant signal component. When isochromatic plus cardiac energy exceeds motion, the step cancels the cardiac signal itself, producing catastrophic collapse. cPACE projects orthogonal to the measured skin-reflectance direction to remove the isochromatic contaminant and recover cardiac direction by eigendecomposition of chrominance covariance (Fig-7). Eigendecomposition selects the dominant direction rather than cancelling it, so cardiac content is preserved. On a Sierra Leone subject, POS retains 1% of the cardiac energy cPACE recovers (Fig-8).

Prior methods assume additive noise, but breathing and speech deform the same vascular bed as the heartbeat, making interference multiplicative and coupled. PRISM models this coupling through homodyne envelope normalization26, respiratory-harmonic rejection, and cross-ROI phase consensus. I evaluated on two datasets27-28 where every rPPG method has struggled, using mean-absolute-error (MAE) on heart rate and my proposed cross-ROI phase-locking-value (PLV) metric11(Fig-9).

For pulse oximetry, the real culprit is wavelength-dependent photon path lengths from melanin absorption. Existing models volume-average absorption29, but melanosomes are discrete particles that concentrate the photon field. I recognized that photon diffusion30 is mathematically equivalent to electrostatics13, enabling Bruggeman effective-medium-theory (EMT)31 to predict a 36% fluence enhancement at 660 nm. I derived a lossy-boundary treatment12 for the 60μm epidermis (too thin to support diffusion) and simulated a homogenized finger via Monte-Carlo32 to compute differential-path-length factors (Fig-10).

Heart rate: cPACE+PRISM achieves equity (Fig-11). On four cohorts spanning skin types and motion conditions, cPACE+PRISM is clinically accurate on all, with no training, no per-subject tuning, and no outlier removal. POS collapses on dark skin; CHROM fails everywhere, locking to a respiratory harmonic. On UBFC-Phys-T2, where every previous method has failed (best achieved MAE = 9.8 BPM27), cPACE+PRISM achieves MAE = 2.8 BPM. Demographic bias in rPPG is not a noise problem; it is a geometry problem.

Phase tells the complete story (Fig-12). Cross-ROI phase locking value (PLV) measures whether cardiac phase is consistent across facial regions. POS collapses from PLV 0.91 on light skin to 0.25–0.27 on dark; CHROM is poor everywhere (<0.25); cPACE maintains 0.61–0.88. Kruskal-Wallis test shows cPACE's PLV does not vary with skin tone (p = 0.64) while POS degrades systematically (p = 4.5×10-12).

Why cPACE+PRISM works everywhere (Fig-13). Component ablation43 reveals what each piece does. cPACE cleans the chrominance subspace by projecting out the isochromatic signal, leaving chromatic cardiac plus uncorrelated noise. It cannot remove motion projected into this plane. PRISM does that: Hilbert envelope normalization and cross-ROI coherence gating recover cardiac signal from multiplicative noise. POS-eig replaces POS's single step of direction finding and noise-cancellation with eigenvector selection, but stays in the [1,1,1] subspace contaminated by isochromatic leakage.

POS is mathematically unstable as it cancels whichever direction dominates variance, including the cardiac signal when isochromatic leakage aligns with it. POS+PRISM improves ideal recordings but stays broken in motion and on dark skin. PRISM cannot recover what has been cancelled. POS-eig+PRISM rescues dark skin at rest (1.2–1.9 BPM) when motion is minimal, but its PLV stays well below cPACE: on CMU India, identical 1.9 BPM MAE but 46% lower PLV. Matched MAE hides degraded signal quality. Under speech (5.5 BPM), motion coupled to isochromatic signal floods the contaminated [1,1,1] plane. cPACE alone is noisy (11.7–33.1 BPM). Only cPACE+PRISM is the principled approach, achieving <3 BPM on every cohort (p < 0.05 on all non-resting cohorts).

SpO₂: predicting a 35-year-old known bias (Fig-14). My EMT + lossy-boundary model predicts demographic SpO₂ bias across 7 independent clinical studies (1990–2025) using only published optical properties: zero free parameters. The lossy boundary accounts for 70% of the correction; EMT refines the remaining 30%. No existing model includes either. Previous Monte Carlo predicts 0%44 or >10%45 bias: neither matches clinical observations. This is the first physics model to predict clinical SpO₂ bias.

TRIOS: three unknowns, three wavelengths (Fig-15). Two-wavelength oximetry is fundamentally underdetermined: arterial oxygen-saturation (SaO₂), melanin, and finger geometry cannot be recovered from two measurements. A third wavelength near the isosbestic point33 collapses bias from ~2.4% to <0.15% across melanin fractions 10–30%. β ≈ 1.47 is a universal correction constant and robust to ±5 nm LED drift (worst case ±0.16%, an order below FDA ±1.5%). No per-device calibration is required.

This work reveals that demographic bias in pulse-oximetry and rPPG is structural, rooted in incomplete physics and not insufficient data.

For years, the field optimized SNR19, but SNR captures strength, not fidelity11. CHROM achieves the highest SNR of any method tested (7.5 dB) yet the worst signal fidelity (PLV = 0.21) (Fig-16). Optimizing strength without fidelity amplifies a corrupted message.

cPACE+PRISM, built from first principles, achieves <3 BPM MAE on every dataset, condition, and skin tone with no training, no calibration, and no subject exclusion; something no existing method has demonstrated47 (Fig-17).

For pulse oximetry, Dr. Aoyagi proposed adding more wavelengths to recover each unknown independently48. My EMT model reveals that melanin and finger geometry couple multiplicatively into a single nuisance, so only one extra wavelength is needed, not two. TRIOS corrects bias from 2.3% to <0.15% with three wavelengths and one universal constant (Fig-18). I hope this work has begun to answer his call14.

Without theory, the field has been searching in the dark. The FDA's 15% dark-skin calibration redistributes bias8, it doesn't remove it. Monte-Carlo studies give wildly different predictions, none matching clinical data (Fig-19)44-46. Lasers, recently proposed49, scatter ±4.4% with sample size of 9. My model predicts an additional 1.6% bias from thermal drift, sitting inside their confidence interval (Fig-20).

I offer the TRIOS intellectual property for $1 to any manufacturer willing to test it. If the physics is right, a 35-year problem gets solved. If wrong, I want to understand why.

Kinetix Lab has invested $250,000 to commercialize this work (Fig-21). A multi-wavelength non-contact co-oximeter integrating cPACE with TRIOS has been built and is being tested.

EMT has never been applied to tissue optics. Preliminary results in two unrelated fields, fundus oximetry52(Fig-22) and optical coherence tomography53(Fig-23), show EMT predicting measured discrepancies that no existing model has explained. This suggests EMT may be a powerful general framework for scattering media.

Beyond health, cPACE's eigenvector framework removes chromatic atmospheric noise from exoplanet transit photometry54 using only the target star's own multi-band data with no external reference needed (Fig-24).

I thank Prof. V. Lakshminarayanan (School of Optometry, University of Waterloo) for guiding me on tissue optics, and projection mathematics. I thank Prof. S. Saini (ECE, University of Waterloo) for teaching me EMT, the cross-disciplinary bridge that made the SpO2 model possible. Both helped me learn how to communicate rigorous science in writing.

I am grateful to the creators of the UBFC-Phys27 and CMU-rPPG28 datasets for making diverse public data available. Without open data, this work would not exist.

I thank Dr. Martin Tobin, whose decades of questioning pulse oximetry bias gave me the conviction that this problem was real and worth pursuing.

I thank my school teachers for flexibility, and WWSEF volunteers for science fair support.

Dedicated to Dr. Takuo Aoyagi (1936–2020), who said: 'Without a theory, there are limits to what devices can do'14. I hope this work answers his call.

References 11-12 are the Journal papers I have submitted and are under peer-review. Reference 13 is a Journal paper under preparation.

W. Verkruysse, L. O. Svaasand, and J. S. Nelson, "Remote plethysmographic imaging using ambient light," Opt. Express, vol. 16, no. 26, pp. 21434-21445, 2008. https://doi.org/10.1364/OE.16.021434

G. Kaur, "SynaptiQ: Detecting hospital-induced delirium," International Science and Engineering Fair, 2025. [Online]. Available: https://isef.net/project/ebed029-synaptiq-detecting-hospital-induced-delirium . Accessed April 16, 2026.

E. M. Nowara, D. McDuff, and A. Veeraraghavan, "A meta-analysis of the impact of skin type and gender on non-contact photoplethysmography measurements," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. Workshops (CVPRW), Seattle, WA, USA, 2020, pp. 1–8.

A. Jubran, and M.J. Tobin, “Reliability of pulse oximetry in titrating supplemental oxygen therapy in ventilator-dependent patients,” Elsevier Chest, 97, 6, pp. 1420-1425, 1990. https://doi.org/10.1378/chest.97.6.1420

M. W. Sjoding, R. P. Dickson, T. J. Iwashyna, S. E. Gay, and T. S. Valley, "Racial bias in pulse oximetry measurement," New England Journal of Medicine, vol. 383, no. 25, pp. 2477–2478, Dec. 2020, doi: https://doi.org/10.1056/NEJMc2029240.

M. J. Tobin and A. Jubran, "Inaccuracy of pulse oximetry in darker-skinned patients is unchanged across 32 years," Eur. Respir. J., vol. 59, no. 6, art. no. 2200520, Jun. 2022, doi: https://doi.org/10.1183/13993003.00520-2022.

A. I. Wong, M. Charpignon, H. Kim, C. Josef, A. A. H. de Hond, J. E. Fojas, A. Tabaie, X. Liu, E. Mireles-Cabodevila, L. Carvalho, R. Kamaleswaran, R. H. Madushani, A. Adhikari, A. Holder, E. W. Steyerberg, T. G. Buchman, M. S. D. Lyons, and L. A. Celi, "Analysis of discrepancies between pulse oximetry and arterial oxygen saturation measurements by race and ethnicity and association with organ dysfunction and mortality," JAMA Network Open, vol. 4, no. 11, p. e2131674, Nov. 2021. doi: https://doi.org/10.1001/jamanetworkopen.2021.31674

U.S. Food and Drug Administration, "Pulse oximeters for medical purposes — Non-clinical and clinical performance testing, labeling, and premarket submission recommendations: Draft guidance for industry and FDA staff," Jan. 7, 2025. [Online]. Available: https://www.fda.gov/media/72470/download

T. Rodriguez, "Pulse oximeter accuracy and racial bias," interview with M. J. Tobin, The Cardiology Advisor, Mar. 2, 2025. [Online]. Available: https://www.thecardiologyadvisor.com/features/pulse-oximeter-accuracy-racial-bias/ Accessed April 16, 2026

G. Boccignone, V. Cuculo, A. D'Amelio, G. Grossi, R. Lanzarotti, and S. Patania, "Remote respiration measurement with RGB cameras: A review and benchmark," ACM Comput. Surv., vol. 58, no. 5, art. no. 114, Nov. 2025, doi: https://doi.org/10.1145/3771763.

G. Kaur and S. S. Saini, "Beyond amplitude: Phase fidelity metrics reveal hidden quality failures in remote photoplethysmography," IOP Physiological Measurement, under peer-review, Apr. 2026.

G. Kaur, V. Lakshminarayanan, and S. S. Saini, "Chrominance phase-aware cardiac eigenprojection (cPACE) for demographically robust remote photoplethysmography," Biomedical Optics Express, under peer-review, Mar. 2026.

G. Kaur and S. S. Saini, "Effective medium theory for skin-pigmentation bias in pulse oximetry: Mechanism, prediction and three-wavelength correction," Biomedical Optics Express, in preparation for submission, May 2026.

K. Miyasaka, S. Shelley, S. Takahashi, H. Kubota, K. Shimada, S. Wei, H. Ogino, H. Ishimura, K. Tohei, Y. Yamamoto, T. Matsuzawa, N. Kobayashi, and M. Fuse, "Tribute to Dr. Takuo Aoyagi, inventor of pulse oximetry," Journal of Anesthesia, vol. 35, no. 5, pp. 671–709, Oct. 2021. [Online]. Available: https://doi.org/10.1007/s00540-021-02967-z

C. Dall'Ora, P. Griffiths, J. Hope, J. Briggs, J. Jeremy, S. Gerry, and O. C. Redfern, "How long do nursing staff take to measure and record patients' vital signs observations in hospital? A time-and-motion study," Int. J. Nurs. Stud., vol. 118, art. no. 103921, Jun. 2021, doi: https://doi.org/10.1016/j.ijnurstu.2021.103921.

U.S. Bureau of Labor Statistics, "Occupational Employment and Wages, May 2024: 29-1141 Registered Nurses," U.S. Department of Labor, Washington, DC, USA, Apr. 2025. [Online]. Available: https://www.bls.gov/oes/2024/may/oes291141.htm

C. Dall'Ora, P. Griffiths, O. Redfern, A. Recio-Saucedo, P. Meredith, and J. Ball, "Nurses' 12-hour shifts and missed or delayed vital signs observations on hospital wards: retrospective observational study," BMJ Open, vol. 9, no. 1, art. no. e024778, Feb. 2019, doi: https://doi.org/10.1136/bmjopen-2018-024778.

G. de Haan and V. Jeanne, "Robust pulse rate from chrominance-based rPPG," IEEE Trans. Biomed. Eng., vol. 60, no. 10, pp. 2878-2886, 2013. https://doi.org/10.1109/TBME.2013.2266196

W. Wang, A. C. den Brinker, S. Stuijk, and G. de Haan, "Algorithmic principles of remote PPG," IEEE Trans. Biomed. Eng., vol. 64, no. 7, pp. 1479-1491, 2017. https://doi.org/10.1109/TBME.2016.2609282

W. Chen and D. McDuff, "DeepPhys: Video-based physiological measurement using convolutional attention networks," Proc. ECCV, pp. 349–365, 2018. https://doi.org/10.48550/arXiv.1805.07888

Z. Yu, X. Li, and G. Zhao, "Remote photoplethysmograph signal measurement from facial videos using spatio-temporal networks," in Proc. British Machine Vision Conference (BMVC), Cardiff, UK, Sep. 2019. [Online]. Available: https://arxiv.org/abs/1905.02419

J. Comas, A. Alomar, A. Ruiz, and F. Sukno, "PhysFlow: Skin tone transfer for remote heart rate estimation through conditional normalizing flows," in Proc. British Machine Vision Conference (BMVC), Glasgow, UK, Nov. 2024. [Online]. Available: https://doi.org/10.48550/arXiv.2407.21519

J. W. Severinghaus, "Takuo Aoyagi: Discovery of pulse oximetry," Anesthesia & Analgesia, vol. 105, no. 6S Suppl., pp. S1–S4, Dec. 2007. [Online]. Available: https://doi.org/10.1213/01.ane.0000269514.31660.09

S. Gupta and N. Aitken, "COVID-19 mortality among racialized populations in Canada and its association with income," StatCan COVID-19: Data to Insights for a Better Canada, Statistics Canada, Catalogue no. 45-28-0001, Aug. 30, 2022. [Online]. Available: https://www150.statcan.gc.ca/n1/pub/45-28-0001/2022001/article/00010-eng.htm

W. Qian, D. Guo, J. Zhou, B. Zou, Z. Yu, and M. Wang, "FreqPhys: Repurposing implicit physiological frequency prior for robust remote photoplethysmography," arXiv preprint arXiv:2604.00534, Apr. 2026. [Online]. Available: https://doi.org/10.48550/arXiv.2604.00534

A. V. Oppenheim and R. W. Schafer, Discrete-Time Signal Processing, 3rd ed. Upper Saddle River, NJ, USA: Prentice Hall, 2010.

R.M. Sabour, Y. Benezeth, P. De Oliveira, J. Chappe, and F. Yang, "UBFC-Phys: A multimodal database for psychophysiological studies of social stress," IEEE Transactions on Affective Computing, vol. 14, no. 1, pp. 622–636, 2023. https://doi.org/10.1109/TAFFC.2021.3056960

A. Dasari, S. K. A. Prakash, L. A. Jeni, and C. S. Tucker, “Evaluation of biases in remote photoplethysmography methods." npj Digital Medicine, 4, no. 1, 91, 2021. https://doi.org/10.1038/s41746-021-00462-z

R. Al-Halawani, I. M. Charlton, M. Qassem, and P. A. Kyriacou, "Semianalytical model based analysis of sources of error in pulse oximetry," J. Biomed. Opt., vol. 29, no. 5, Art. no. 057001, May 2024, doi: https://doi.org/10.1117/1.JBO.29.5.057001.

L. V. Wang and H.-i. Wu, Biomedical Optics: Principles and Imaging. Hoboken, NJ, USA: Wiley, 2007.

D. A. G. Bruggeman, "Berechnung verschiedener physikalischer Konstanten von heterogenen Substanzen. I. Dielektrizitätskonstanten und Leitfähigkeiten der Mischkörper aus isotropen Substanzen [Calculation of various physical constants of heterogeneous substances. I. Dielectric constants and conductivities of mixed bodies of isotropic substances]," Ann. Phys., vol. 416, no. 7, pp. 636–664, 1935, https://doi.org/10.1002/andp.19354160705

M. S. Arefin, A. P. Dumont, and C. A. Patil, "Monte Carlo based simulations of racial bias in pulse oximetry," Proc. SPIE, vol. 11951, Art. no. 1195103, Mar. 2022, doi: https://doi.org/10.1117/12.2610483.

S. L. Jacques, "Optical properties of biological tissues: A review," Phys. Med. Biol., vol. 58, no. 11, pp. R37–R61, Jun. 2013, doi: https://doi.org/10.1088/0031-9155/58/11/R37.

S. A. Prahl, "Tabulated molar extinction coefficient for hemoglobin in water," Oregon Medical Laser Center, Portland, OR, USA, 1999. [Online]. Available: https://omlc.org/spectra/hemoglobin/summary.html

G. M. Hale and M. R. Querry, "Optical constants of water in the 200-nm to 200-μm wavelength region," Appl. Opt., vol. 12, no. 3, pp. 555–563, Mar. 1973, doi: https://doi.org/10.1364/AO.12.000555.

J. Crank, The Mathematics of Diffusion, 2nd ed. Oxford, U.K.: Clarendon Press, 1975.

D. J. Griffiths, Introduction to Electrodynamics, 4th ed. Cambridge, U.K.: Cambridge University Press, 2017.

P. E. Bickler, J. R. Feiner, and J. W. Severinghaus, “Effects of skin pigmentation on pulse oximeter accuracy at low saturation,” Anesthesiology, 102, 4, pp. 715-719, 2005. https://doi.org/10.1097/00000542-200504000-00004

G. Leeb, I. Auchus, T. Law, P. Bickler, J. Feiner, S. Hashi, E. Monk, E. Igaga, M. Bernstein, Y.-C. Chou, C. Hughes, D. Schornack, J. Lester, K. Moore Jr., O. Okunlola, J. Fernandez, S. Shmuylovich, and M. S. Lipnick, "The performance of 11 fingertip pulse oximeters during hypoxemia in healthy human participants with varied, quantified skin pigment," EBioMedicine, vol. 102, Art. no. 105051, Apr. 2024. https://doi.org/10.1016/j.ebiom.2024.105051

J. R. Starnes, W. Welch, C. C. Henderson, S. Hudson, S. Risney, G. T. Nicholson, T. P. Doyle, D. R. Janssen, B. P. Londergan, D. A. Parra, J. C. Slaughter, M. H. Aliyu, J. A. Graves, and J. H. Soslow, "Pulse oximetry and skin tone in children," N. Engl. J. Med., vol. 392, no. 10, pp. 1033–1034, Mar. 2025, doi: https://doi.org/10.1056/NEJMc2414937.

A. Fawzy, T. D. Wu, K. Wang, M. L. Robinson, J. Farha, A. Bradke, S. H. Golden, Y. Xu, and B. T. Garibaldi, "Racial and ethnic discrepancy in pulse oximetry and delayed identification of treatment eligibility among patients with COVID-19," JAMA Internal Medicine, vol. 182, no. 7, pp. 730–738, Jul. 2022.  https://doi.org/10.1001/jamainternmed.2022.1906

C. J. Crooks, J. West, J. R. Morling, M. Simmonds, I. Juurlink, S. Briggs, S. Cruickshank, S. Hammond-Pears, D. Shaw, T. R. Card, and A. W. Fogarty, "Pulse oximeter measurements vary across ethnic groups: An observational study in patients with COVID-19," Eur. Respir. J., vol. 59, no. 4, Art. no. 2103246, Apr. 2022, doi: 10.1183/13993003.03246-2021.

A. Newell, "A tutorial on speech understanding systems," in Speech Recognition: Invited Papers Presented at the 1974 IEEE Symposium, D. R. Reddy, Ed. New York, NY, USA: Academic Press, 1975, pp. 3–54.

S. K. Narayanaswamy, C. Liu, R. Correia, B. R. Hayes-Gill, and S. P. Morgan, "Exploring the bias: how skin color influences oxygen saturation readings via Monte Carlo simulations," J. Biomed. Opt., vol. 29, no. S3, art. no. S33308, Aug. 2024, doi: https://doi.org/10.1117/1.JBO.29.S3.S33308.

R. Al-Halawani, M. Qassem, and P. A. Kyriacou, "Monte Carlo simulation of the effect of melanin concentration on light–tissue interactions for transmittance pulse oximetry measurement," J. Biomed. Opt., vol. 29, no. S3, art. no. S33305, Aug. 2024, doi: https://doi.org/10.1117/1.JBO.29.S3.S33305.

G. Blaney, A. Sassaroli, and S. Fantini, "Critical analysis of the relationship between arterial saturation and the ratio-of-ratios used in pulse oximetry," J. Biomed. Opt., vol. 29, no. S3, art. no. S33313, Nov. 2024, doi: https://doi.org/10.1117/1.JBO.29.S3.S33313.

X. Liu, G. Narayanswamy, A. Paruchuri, X. Zhang, J. Tang, Y. Zhang, R. Sengupta, S. Patel, Y. Wang, and D. McDuff, "rPPG-Toolbox: Deep remote PPG toolbox," in Proc. 37th Conf. Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track, New Orleans, LA, USA, Dec. 2023. [Online]. Available: https://arxiv.org/abs/2210.00716

T. Aoyagi, M. Fuse, N. Kobayashi, K. Machida, and K. Miyasaka, "Multiwavelength pulse oximetry: Theory for the future," Anesth. Analg., vol. 105, no. 6 Suppl., pp. S53–S58, Dec. 2007, doi: https://doi.org/10.1213/01.ane.0000268716.07255.2b.

J. A. Pologe, N. K. You, M. Blumstein, K. L. Snyder, and W. W. Hay, "Laser-based pulse oximetry eliminates pigmentation effects on oxygen saturation measurements: A pilot study," PLoS One, vol. 20, no. 10, art. no. e0333109, Oct. 2025, doi: https://doi.org/10.1371/journal.pone.0333109.

M. Elgendi, I. Martinelli, and C. Menon, "Optimal signal quality index for remote photoplethysmogram sensing," npj Biosensing, vol. 1, no. 1, art. no. 5, Jun. 2024, doi: https://doi.org/10.1038/s44328-024-00002-1

M. Bondarenko, M. Menon, and M. Elgendi et al., "Demographic bias in public remote photoplethysmography datasets," npj Digital Med., vol. 8, art. no. 479, Oct. 2025, doi: https://doi.org/10.1038/s41746-025-01973-9.

A. K. Garg, D. Knight, L. Lando, and D. L. Chao, "Advances in retinal oximetry," Transl. Vis. Sci. Technol., vol. 10, no. 2, art. no. 5, Feb. 2021, doi: https://doi.org/10.1167/tvst.10.2.5.

R. F. Spaide, H. Koizumi, and M. C. Pozzoni, "Enhanced depth imaging spectral-domain optical coherence tomography," Am. J. Ophthalmol., vol. 146, no. 4, pp. 496–500, Oct. 2008, doi: https://doi.org/10.1016/j.ajo.2008.05.032.

D. Charbonneau, T. M. Brown, D. W. Latham, and M. Mayor, "Detection of planetary transits across a Sun-like star," Astrophys. J. Lett., vol. 529, no. 1, pp. L45–L48, Jan. 2000, doi: https://doi.org/10.1086/312457.

M. Hammer, W. Vilser, T. Riemer, and D. Schweitzer, "Retinal vessel oximetry — calibration, compensation for vessel diameter and fundus pigmentation, and reproducibility," J. Biomed. Opt., vol. 13, no. 5, art. no. 054015, Sep.–Oct. 2008, doi: https://doi.org/10.1117/1.2976032.

H. A. Khairy, H. A. M. Elabshihy, and M. A. Zaky, "Choroidal change assessment with enhanced depth imaging optical coherence tomography in myopic choroidal neovascularization," Menoufia Med. J., vol. 32, no. 2, 2019. [Online]. Available: https://www.menoufia-med-j.com/journal/vol32/iss2/

M. A. Zouache, C. D. Faust, V. Silvestri, S. Akafo, S. Lartey, R. Mehta, J. Carroll, G. Silvestri, G. S. Hageman, and W. M. Amoaku, "Retinal and choroidal thickness in an indigenous population from Ghana: Comparison with individuals with European or African ancestry," Ophthalmol. Sci., vol. 4, no. 2, art. no. 100386, Aug. 2023, doi: https://doi.org/10.1016/j.xops.2023.100386.

A. Sanchez-Cano, E. Orduna, F. Segura, C. Lopez, N. Cuenca, E. Abecia, and I. Pinilla, "Choroidal thickness and volume in healthy young white adults and the relationships between them and axial length, ametropy and sex," Am. J. Ophthalmol., vol. 158, no. 3, pp. 574–583.e1, Sep. 2014, doi: https://doi.org/10.1016/j.ajo.2014.05.035.