PulseLens

Accessible contactless vital monitoring that democratizes cardiovascular health screening for underserved communities using standard camera sensors.

Comment

Inspiration

Access to routine cardiovascular checkups remains severely unequal across local communities and rural clinics. Dedicated diagnostic hardware—such as clinical pulse oximeters, ECG monitors, and photoplethysmography (PPG) sensors—is often cost-prohibitive, bulky, or simply unavailable in overburdened healthcare centers. We were inspired to bridge this healthcare disparity by transforming standard consumer webcams and smartphone cameras into contactless, real-time diagnostic tools for underserved populations.

What it does

PulseLens is a non-invasive, contactless physiological vital monitoring platform. By analyzing microscopic color fluctuations in facial skin caused by blood flow through the cardiac cycle (remote photoplethysmography, or rPPG), PulseLens calculates:

  • Real-time Heart Rate (BPM)
  • Pulse Rate Variability (PRV) insights
  • Signal waveform visualization and health trend monitoring

Anyone with a standard camera-enabled device can initiate a quick screening session, making vital sign checks completely frictionless and accessible without specialized medical peripherals.

How we built it

  • Computer Vision & Facial Tracking: Detected and tracked facial regions of interest (ROI), specifically forehead and cheek areas where capillary blood flow is most pronounced, while minimizing ambient motion artifacts.
  • Signal Extraction & Processing: Isolated the green color channel—which exhibits the highest absorption contrast for hemoglobin—and extracted the raw photoplethysmographic signal over time.
  • Filtering & Frequency Analysis: Applied temporal bandpass filtering (0.7 Hz to 3.5 Hz, corresponding to 42–210 BPM) and Fast Fourier Transform (FFT) / peak detection algorithms to extract clean cardiac frequencies from optical noise.
  • Interface & Deployment: Wrapped the pipeline into an intuitive, lightweight web interface allowing instant one-click vital checks.

Challenges we faced

  • Environmental Lighting & Noise: Variations in ambient room illumination and auto-exposure adjustments caused baseline drift in the optical signal. We had to tune filtering and normalization thresholds to separate cardiovascular pulses from ambient flicker.
  • Motion Artifacts: Even minor head movements introduce significant noise to optical ROI tracking. Developing dynamic ROI stabilization was crucial to maintaining reliable readings.
  • Latency Optimization: Processing video frames in real time while maintaining smooth rendering and fast signal convergence required careful optimization of computational loops.

Accomplishments that we're proud of

  • Successfully extracting clean, identifiable cardiac waveforms purely through optical camera feeds without any contact hardware.
  • Delivering a functional, lightweight tool capable of running on modest consumer hardware.
  • Aligning engineering craft directly with a meaningful civic and community health mission.

What we learned

  • The mathematical foundations of digital signal processing, Fourier analysis, and color-space transformations in physiological computing.
  • Practical techniques for stabilizing computer vision pipelines against real-world environmental disturbances.
  • How software-driven health tech can dramatically reduce barriers to preventative screening in public health contexts.

What's next for PulseLens

  • Respiration Rate & SpO2 Estimation: Expanding optical analysis to approximate respiratory rate and oxygen saturation.
  • Edge & Mobile Optimization: Optimizing the model pipeline to run seamlessly on low-end mobile devices and offline community health kiosks.
  • Clinical Validation & Telehealth APIs: Partnering with regional clinics to calibrate algorithm accuracy and providing simple webhooks for remote telemedicine integration.

Built With

Share this project:

Updates