Researchers created a fiber-compatible, all-optical biomimetic spike encoder enabling rapid, precise conversion of medical signals for neuromorphic analysis.
Key Details
- 1All-optical encoder uses optically injected Fabry-Perot laser neurons for spike rate >2.0 GHz and ≈5-bit precision.
- 2Device operates at standard 1,550 nm fiber band, compatible with current telecom infrastructure.
- 3Spiking neural network integration achieves 94.1% MNIST image classification accuracy.
- 4Proof-of-concept demonstrated on chest X-ray, ECG arrhythmia, and heart-sound data for healthcare diagnosis.
- 5Each spike consumes only 0.5 pJ, outperforming many electrical/optical alternatives in energy efficiency.
- 6Encoding remains robust after 10 km transmission over standard single-mode fiber.
Why It Matters
This innovation allows direct, scalable integration of neuromorphic AI into existing healthcare and communication networks, paving the way for real-time, low-power medical data processing and diagnostics.
Source
EurekAlert
Related News
•EurekAlert
Sydney Team Reveals Fractal Limits to Predictability in AI Training
University of Sydney researchers discover that deep learning models may remain unpredictable due to deep-rooted mathematical properties, even under controlled training conditions.
•EurekAlert
Machine Learning Tool Maps Alzheimer’s Risk on fMRI Spectrum
Researchers developed an AI framework using resting-state fMRI to assess Alzheimer's risk along a continuum from SuperAgers to AD patients.
•EurekAlert
Mayo Clinic Unveils AI-Enabled Clinical Decision Support Platform
Mayo Clinic Platform launches an AI-enabled Clinical Decision Support solution to streamline clinician workflows and provide evidence-based, individualized patient care.