Shenzhen University Builds Photonic AI Diagnostic Platform With 246-Fold Efficiency Gain
Updated
Updated · EIN Presswire · Jun 17
Shenzhen University Builds Photonic AI Diagnostic Platform With 246-Fold Efficiency Gain
1 articles · Updated · EIN Presswire · Jun 17
Summary
Shenzhen University researchers and industry partners built an all-fiber photonic neural network for medical diagnosis, replacing electronic processing with light-based computing to cut power use and speed image analysis.
95.0% accuracy and 97.6% specificity were reported in liver cancer diagnosis from 3,348 CT studies, while a retinal detachment test used 80 ultrasound images; the team said performance matched experienced radiologists.
0.8 milliseconds were needed to process one liver CT study on the photonic system versus 85 ms on an NVIDIA A100 GPU, and energy per operation fell to 0.608 fJ from 150 fJ.
The platform relies on black phosphorus and MoS2 heterostructures integrated onto microfiber knot resonators and RAMZI devices, addressing long-standing optical modulator limits in efficiency, size and fabrication.
The current setup uses only two modulators in a single layer, so the team is targeting a 40-channel wavelength-division multiplexing design and industrial encapsulation to scale toward clinical deployment.
With AI vision already superhuman, can greener hardware finally bridge the gap to clinical practice?
If light-based AI diagnoses cancer in milliseconds, what other impossible problems could it solve?
246x More Efficient: Shenzhen University’s Black Phosphorus Photonic AI Sets New Benchmark for Medical Diagnostics
Overview
A collaborative team led by Professor Han Zhang at Shenzhen University, together with industry partners, has developed a groundbreaking black phosphorus-based all-fiber photonic AI diagnostic platform. Officially published in May 2026, this innovation marks a major advance in sustainable and effective AI, especially for medical applications. The new platform stands out for its ability to deliver exceptional energy efficiency and high diagnostic accuracy. By leveraging photonic technology, it drastically reduces the energy needed for complex AI computations while improving precision in critical tasks like medical diagnosis, paving the way for more reliable and eco-friendly AI solutions.