Label-free optical fundus angiography based on programmable hyperspectral imaging

TY - JOUR

T1 - Label-free optical fundus angiography based on programmable hyperspectral imaging

AU - Ren, Yuetian

AU - Song, Yuning

AU - Yao, Jinchi

AU - Wei, Hongguo

AU - Wang, Yunxin

AU - Luo, Shuaihang

AU - Xu, Wenbin

AU - Joseph, James

AU - Chen, Keren

AU - Chen, Shuo

N1 - Publisher Copyright: © 2025 Elsevier Ltd

PY - 2025/12

Y1 - 2025/12

N2 - Retinal vasculature, as the sole deep vascular network within the human body that can be non-invasively observed, is closely associated with various retinal and systemic disorders. However, conventional fundus imaging techniques often suffer from inadequate contrast or rely on invasive contrast agents to visualize fine retinal vessels, hemorrhages, and exudates, thereby posing great challenges in screening or diagnosing fundus diseases at their early stages. In this study, we report a label-free optical fundus angiography method based on programmable hyperspectral imaging, in which fundus images with high-contrast retinal blood vessels can be captured in a snapshot by exposing fundus tissues under specific spectrally coded illumination. Using a self-developed programmable light source, such specific spectrally coded illumination is multiplexed with specific intensities at different wavelengths under the guidance of spectral post-processing algorithms. This enables the preservation of key spectral bands that can maximize the differences between the retinal blood vessels and other fundus tissues. In such a manner, the proposed method can select the wavelengths of interest during fundus imaging in a snapshot, in contrast to numerical wavelength selection after acquiring the full wavelengths of conventional spectroscopic fundus imaging techniques; thus, the time-consuming acquisition of a three-dimensional spectral dataset and spectral post-processing can be avoided. Through evaluations on a model eye and ex vivo porcine eyes, the proposed label-free optical fundus angiography method shows significant enhancement in the image contrast of retinal blood vessels compared to conventional fundus imaging, thus is expected to benefit early and more precise diagnosis of fundus diseases.

AB - Retinal vasculature, as the sole deep vascular network within the human body that can be non-invasively observed, is closely associated with various retinal and systemic disorders. However, conventional fundus imaging techniques often suffer from inadequate contrast or rely on invasive contrast agents to visualize fine retinal vessels, hemorrhages, and exudates, thereby posing great challenges in screening or diagnosing fundus diseases at their early stages. In this study, we report a label-free optical fundus angiography method based on programmable hyperspectral imaging, in which fundus images with high-contrast retinal blood vessels can be captured in a snapshot by exposing fundus tissues under specific spectrally coded illumination. Using a self-developed programmable light source, such specific spectrally coded illumination is multiplexed with specific intensities at different wavelengths under the guidance of spectral post-processing algorithms. This enables the preservation of key spectral bands that can maximize the differences between the retinal blood vessels and other fundus tissues. In such a manner, the proposed method can select the wavelengths of interest during fundus imaging in a snapshot, in contrast to numerical wavelength selection after acquiring the full wavelengths of conventional spectroscopic fundus imaging techniques; thus, the time-consuming acquisition of a three-dimensional spectral dataset and spectral post-processing can be avoided. Through evaluations on a model eye and ex vivo porcine eyes, the proposed label-free optical fundus angiography method shows significant enhancement in the image contrast of retinal blood vessels compared to conventional fundus imaging, thus is expected to benefit early and more precise diagnosis of fundus diseases.

UR - https://www.scopus.com/pages/publications/105008103841

U2 - 10.1016/j.optlastec.2025.113389

DO - 10.1016/j.optlastec.2025.113389

M3 - Article

AN - SCOPUS:105008103841

SN - 0030-3992

VL - 191

JO - Optics and Laser Technology

JF - Optics and Laser Technology

M1 - 113389

ER -