English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In optical coherence tomography (OCT) volumes of retina, the sequentialacquisition of the individual slices makes this modality prone to motionartifacts, misalignments between adjacent slices being the most noticeable. Anydistortion in OCT volumes can bias structural analysis and influence theoutcome of longitudinal studies. On the other hand, presence of speckle noisethat is characteristic of this imaging modality, leads to inaccuracies whentraditional registration techniques are employed. Also, the lack of awell-defined ground truth makes supervised deep-learning techniques ill-posedto tackle the problem. In this paper, we tackle these issues by using deepreinforcement learning to correct inter-frame movements in an unsupervisedmanner. Specifically, we use dueling deep Q-network to train an artificialagent to find the optimal policy, i.e. a sequence of actions, that bestimproves the alignment by maximizing the sum of reward signals. Instead ofrelying on the ground-truth of transformation parameters to guide the rewardingsystem, for the first time, we use a combination of intensity based imagesimilarity metrics. Further, to avoid the agent bias towards speckle noise, weensure the agent can see retinal layers as part of the interacting environment.For quantitative evaluation, we simulate the eye movement artifacts by applying2D rigid transformations on individual B-scans. The proposed model achieves anaverage of 0.985 and 0.914 for normalized mutual information and correlationcoefficient, respectively. We also compare our model with elastix intensitybased medical image registration approach, where significant improvement isachieved by our model for both noisy and denoised volumes.

Dueling Deep Q-Network for Unsupervised Inter-frame Eye Movement Correction in Optical Coherence Tomography Volumes