Deep Learning Enabled Scalable Calibration of a Dynamically Deformed Multimode Fiber

Fan, Pengfei; Wang, Yufei; Ruddlesden, Michael; Wang, Xuechun; Thaha, Mohamed A.; Sun, Jiasong; Zuo, Chao; Su, Lei

doi:10.1002/adpr.202100304

Cited by 11 publications

(2 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this method requires high computational time. Recent advances in AI show great promise for real-time image processing and reconstruction through multi-mode fibre [11][12][13] , however these models are limited to accommodating fibre perturbations that affect the amplitude of the images rather than learning the dynamic physical change of TMs.…”

Section: Introductionmentioning

confidence: 99%

Reflection-mode transmission matrix reconstruction using neural networks in optical fiber imaging systems

Zheng,

Gordon

2024

AI and Optical Data Sciences V

View full text Add to dashboard Cite

Needle-thin optical fibre imaging systems using multimode fibre show considerable potential for facilitating advanced medical endoscopes that can capture high-resolution images in challenging regions of the body, such as the brain or blood vessels. However, these systems experience significant optical distortion whenever the fibre is disturbed. To address this, it is crucial to calibrate the fibre transmission matrix (TM) in vivo immediately before conducting the imaging process since TM is highly sensitive to temperature variations and bending. We therefore present a reflection-mode TM reconstruction model using U-net based convolutional neural networks with a custom loss function used for arbitrary global phase compensation, which reduced computational time to ~1s. We demonstrated this model by reconstructing 64 × 64 complex-valued fibre TMs through a reflection-mode optical fibre system and tested by reconstructing widefield images with ≤ 9% image error. We anticipate this neural network-based TM reconstruction model with the custom loss function designed will lead to new AI models that deal with phase information, for example in imaging through optical fibre, holographic imaging and projection, where both phase control and speed are required.

show abstract

Section: Introductionmentioning

confidence: 99%

Reflection-mode transmission matrix reconstruction using neural networks in optical fiber imaging systems

Zheng,

Gordon

2024

AI and Optical Data Sciences V

View full text Add to dashboard Cite

show abstract

“…Recently, deep learning-based approaches have been demonstrated for image transmission through MMFs. − MMF imaging using deep learning typically uses spatial light modulators to generate the object patterns and record the corresponding speckle to establish the dataset for neural network training. Based on trained neural networks, real-time or even ultrafast imaging can be achieved. , Deep learning can also be used to improve the imaging quality through dynamically perturbed MMF. − Traditional methods of MMF anti-perturbation imaging using deep learning usually require collecting a large number of object–speckle pairs under different fiber configurations for neural network training. These methods are usually applicable to specific optical fiber configurations, and the realization of MMF imaging in unknown configurations depends on the data acquisition in hundreds of configurations.…”

Section: Introductionmentioning

confidence: 99%

Anti-perturbation Multimode Fiber Imaging Based on the Active Measurement of the Fiber Configuration

et al. 2023

View full text Add to dashboard Cite

Multimode fiber (MMF) imaging is an emerging field of fiber imaging technology in the last few decades. However, its high sensitivity to dynamic perturbance limits its practical applications. In this study, we propose an anti-perturbation scheme for MMF imaging based on the active measurement of the fiber configuration. We fabricate an imaging device composed of the MMF and fiber Bragg grating array to measure the MMF configuration parameters in real time and record the object–speckle pairs in different configurations for neural network training. Image reconstruction subjected to dynamic perturbations can be realized using deep learning, and the experimental results show that the introduction of fiber configuration parameters can improve the quality of anti-perturbation imaging. In addition, we realize speckle prediction using the configuration parameters and a trained neural network. The predicted speckle can be applied to flexible MMF compressive imaging. Our work proposes a new scheme for flexible MMF imaging and provides an important reference for the practical application of MMF imaging.

show abstract

Single-ended recovery of optical fiber transmission matrices using neural networks

Zheng,

Wright,

Wen

et al. 2023

Commun Phys

View full text Add to dashboard Cite

Ultra-thin multimode optical fiber imaging promises next-generation medical endoscopes reaching high image resolution for deep tissues. However, current technology suffers from severe optical distortion, as the fiber’s calibration is sensitive to bending and temperature and thus requires in vivo re-measurement with access to a single end only. We present a neural network (NN)-based approach to reconstruct the fiber’s transmission matrix (TM) based on multi-wavelength reflection-mode measurements. We train two different NN architectures via a custom loss function insensitive to global phase-degeneracy: a fully connected NN and convolutional U-Net. We reconstruct the 64 × 64 complex-valued fiber TMs through a simulated single-ended optical fiber with ≤ 4% error and cross-validate on experimentally measured TMs, demonstrating both wide-field and confocal scanning image reconstruction with small error. Our TM recovery approach is 4500 times faster, is more robust to fiber perturbation during characterization, and operates with non-square TMs.

show abstract

Deep Learning Enabled Scalable Calibration of a Dynamically Deformed Multimode Fiber

Cited by 11 publications

References 50 publications

Reflection-mode transmission matrix reconstruction using neural networks in optical fiber imaging systems

Reflection-mode transmission matrix reconstruction using neural networks in optical fiber imaging systems

Anti-perturbation Multimode Fiber Imaging Based on the Active Measurement of the Fiber Configuration

Single-ended recovery of optical fiber transmission matrices using neural networks

Contact Info

Product

Resources

About