Fragile-skin compatible design
Aims to reduce placement burden on delicate neonatal skin.
Solution architecture
Solution / Technology
NeoPerfuse combines wavelength-engineered multi-channel optical sensing with an explainable AI-based Signal Quality Index to make neonatal oxygenation and perfusion signals more trustworthy.
System architecture
Step 1
Skin-friendly patch
A neonatal-compatible patch designed for fragile skin, small extremities, and difficult placement.
Step 2
Multi-channel optical signals
Selected channels target SpO₂ baseline, tissue oxygenation, perfusion context, and water/contact effects.
Step 3
Explainable AI-SQI
AI evaluates each channel and time segment to reject artifacts and classify likely failure modes.
Step 4
Signal confidence + reason
The output shows whether the optical signal is trustworthy and why it may be unreliable.
Neonatal-compatible sensor concept
A skin-friendly sensor head or patch designed for very small extremities and fragile neonatal skin. The design objective is fast placement, stable optical coupling, low-burden use, and compatibility with delivery-room and NICU workflows.
Small-extremity form factor
Designed around the constraints of preterm anatomy.
Fast and reproducible placement
Supports delivery-room and NICU workflows under time pressure.
Stable optical coupling
Focused on difficult wet-skin, contact, and motion conditions.
Multispectral optical acquisition
The system acquires optical raw signals across multiple wavelength channels. Additional NIR/SWIR-adjacent channels may support perfusion sensitivity, artifact assessment, and signal-quality discrimination rather than being presented as a broad claim of superior SpO2 accuracy.
Conventional optical channels
Reference-compatible optical acquisition concepts.
Additional multispectral channels
Exploratory wavelength channels for signal-quality discrimination.
Perfusion-sensitive analysis
Designed to evaluate pulsatility and low-perfusion behavior.
Artifact and contact-quality assessment
Aims to make signal reliability more transparent.
AI-based Signal Quality Index
The AI component is designed as clinically explainable trust logic rather than a black-box oxygen-value estimator. It evaluates whether the optical information is interpretable and may classify why a signal is unreliable.
Motion artifact
Movement affecting waveform reliability.
Low perfusion
Weak pulsatility or poor peripheral signal.
Poor sensor contact
Suboptimal optical coupling.
Wet skin
Moisture affecting sensor-skin interface.
Sensor displacement
Patch movement or misalignment.
Insufficient pulsatility
Limited interpretable waveform content.
Confidence-aware clinical output
Confidence-aware output
Signal qualityHigh
Motion likelihoodLow
Signal confidence
High / Medium / Low
Likely failure mode
Motion / Low perfusion / Poor contact / Wet skin / Displacement
Suggested interpretation
Signal interpretable / not interpretable / check contact / prioritize reference monitoring
Collaboration