Deep learning: step forward to high‐resolution in vivo shortwave infrared imaging - Archive ouverte HAL Access content directly
Journal Articles Journal of Biophotonics Year : 2021

Deep learning: step forward to high‐resolution in vivo shortwave infrared imaging

(1) , (2) , (3)
1
2
3

Abstract

In vivo optical imaging in the shortwave infrared windows (1000-1700 nm) has shown a growing interest with major improvement in terms of spatial and temporal resolution in depth down to 4 mm compared to the NIR-I region (700-900 nm). This method can be particularly useful for studies of the growth and development of blood vessels in tumors, in vivo monitoring of pathologies and evaluation of effects of drugs. SWIR signal obtained from vessels passes through tissues and skin and thus, subject to noise and scattering. We demonstrate that the combination of SWIR imaging in the NIR-IIb (1500-1700 nm) region with advanced deep learning image analysis on small animals can provide a non-intrusive deep insight into the morphology of the blood vessels. For demonstration we use neural network IterNet that exploits structural redundancy of the blood vessels (L. Li, et.al., The IEEE WACV, 2020). It can reconstruct the blood vessels structure in high details, thus providing a useful analysis tool for raw SWIR images.
Fichier principal
Vignette du fichier
Deep_learning_of_vessels-2.pdf (2.5 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03365751 , version 1 (05-10-2021)

Identifiers

Cite

Vladimir A Baulin, Yves Usson, Xavier Le Guével. Deep learning: step forward to high‐resolution in vivo shortwave infrared imaging. Journal of Biophotonics, 2021, 14 (7), pp.e202100102. ⟨10.1002/jbio.202100102⟩. ⟨hal-03365751⟩
361 View
13 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More