Agentic AI-powered flexible fiber-bundle endoscopy for high-resolution NIR-II fluorescence imaging in vivo

2026-08-08

AI Summary

This study addresses the limitations of conventional fiber-optic endoscopes in second near-infrared window (NIR-II) imaging—namely, low resolution, honeycomb artifacts, and inter-core crosstalk—which hinder high-fidelity in vivo fluorescence visualization. To overcome these challenges, the authors propose an optics-computation co-design strategy: first, an ultrafine fiber bundle is engineered to suppress optical crosstalk; second, a Guided Agent-based Mixture-of-Experts (GAME) image restoration framework is developed, integrating vision-language models to enable dynamic routing that adaptively tailors reconstruction to diverse biological specimens. This approach surpasses the Nyquist–Shannon sampling limit, achieving a fourfold resolution enhancement. The system demonstrates high-resolution NIR-II endoscopic imaging of murine anatomical structures as well as human gastric tubes and lymphatic systems, thereby advancing clinical translation.

Abstract

Fiber-bundle endoscopy offers a compact and flexible route for clinical fluorescence imaging through natural human orifices, but since its first report in the 1950s, it has remained limited by low spatial resolution, honeycomb artifacts, and inter-core crosstalk. The crosstalk becomes more pronounced at near-infrared-II wavelengths (NIR-II, 1000-3000 nm), a spectral window that offers superior contrast, resolution, and tissue penetration depth for biomedical imaging. Here, we present an AI-powered flexible endoscopy platform that overcomes these constraints through optical-computational co-design: optimizing ultrathin fiber bundles to mitigate crosstalk-induced image blur and enable high-fidelity image transmission across the visible-to-NIR-II spectral range, and developing an Agent-Guided Mixture-of-Experts (GAME) pipeline for honeycomb-artifact removal and image restoration. GAME provides a single restoration entry point for diverse biomedical images acquired with our endoscope, spanning cell, mouse and human samples. It dynamically routes each input to suitable restoration experts via a vision-language model, facilitating image reconstruction with a fourfold resolution improvement beyond the NyquistShannon sampling limit. The utility of our endoscope is demonstrated through in vivo NIR-II imaging of anatomical structures in mice, as well as imaging of the digital micromirror device (DMD)-projected human gastric tube and lymphatic system, paving the way for future clinical translation.

Problem

• fiber-bundle endoscopy
• NIR-II fluorescence imaging
• inter-core crosstalk
• honeycomb artifacts
• spatial resolution

Innovation

• Agentic AI
• fiber-bundle endoscopy
• NIR-II imaging
• optical-computational co-design
• Mixture-of-Experts