Synthetic high-resolution, volumetric and wide field-of-view optical coherence tomography images with generative adversarial networks
Journal
Proceedings of SPIE - The International Society for Optical Engineering
Journal Volume
12632
ISBN
9781510664739
Date Issued
2023-01-01
Author(s)
Chueh, Chuan Bor
Chen, Ting Hao
Li, Yu Yu
Tu, Ming Che
Lee, Cheng Kuang
See, Simon
Abstract
Optical coherence tomography (OCT) has been widely used in many clinical apartments. The trade-off between field-of-view (FOV) and transverse resolution has always become critical. High-resolution (HR) image reconstruction of OCT has the potential to increase the resolution without reducing the FOV. Although it has been widely explored in these years, it is hard to find one-to-one paired high-resolution for reference. The SR methods are used to improve the cross-sectional OCT image only. Therefore, we build a custom-design, wide field-of-view (FOV), multi-scale optical coherence tomography that can allow us to produce excellent high- and low-resolution one-to-one mapping volumetric data set. With the SR methods, we can reduce the acquisition of synthetic high-resolution volumetric images by up to four times.
Subjects
deep learning | GAN | multi-scale imaging | optical coherence tomography | super-resolution
Type
conference paper