Backscattered Light Repairing Method for Underwater Laser Image Based on Improved Generative Adversarial Network

Qingbo, 张清博 Zhang; Xiaohui, 张晓晖 Zhang; Hongwei, 韩宏伟 Han

doi:10.3788/lop56.041004

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is because cCNNs require a large number of parameters to accommodate all potential features in images, which can lead to overfitting, as reported in previous studies ( 52 , 60 ). The finding that ResNet outperformed GAN, however, was surprising to us because GAN has shown improved performances in many conventional image processing tasks including realistic image generation ( 61 ), inpainting ( 62 ), and image repairing ( 63 ). One possible explanation is that the loss functions used by the generator and discriminator here were not designed specifically to minimize the RMSE.…”

Section: Discussionmentioning

confidence: 93%

Mouse brain MR super-resolution using a deep learning network trained with optical imaging data

Liang

Zhang

2023

Front. Radiol.

View full text Add to dashboard Cite

IntroductionThe resolution of magnetic resonance imaging is often limited at the millimeter level due to its inherent signal-to-noise disadvantage compared to other imaging modalities. Super-resolution (SR) of MRI data aims to enhance its resolution and diagnostic value. While deep learning-based SR has shown potential, its applications in MRI remain limited, especially for preclinical MRI, where large high-resolution MRI datasets for training are often lacking.MethodsIn this study, we first used high-resolution mouse brain auto-fluorescence (AF) data acquired using serial two-photon tomography (STPT) to examine the performance of deep learning-based SR for mouse brain images.ResultsWe found that the best SR performance was obtained when the resolutions of training and target data were matched. We then applied the network trained using AF data to MRI data of the mouse brain, and found that the performance of the SR network depended on the tissue contrast presented in the MRI data. Using transfer learning and a limited set of high-resolution mouse brain MRI data, we were able to fine-tune the initial network trained using AF to enhance the resolution of MRI data.DiscussionOur results suggest that deep learning SR networks trained using high-resolution data of a different modality can be applied to MRI data after transfer learning.

show abstract