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

The performance of many medical image analysis tasks are strongly associatedwith image data quality. When developing modern deep learning algorithms,rather than relying on subjective (human-based) image quality assessment (IQA),task amenability potentially provides an objective measure of task-specificimage quality. To predict task amenability, an IQA agent is trained usingreinforcement learning (RL) with a simultaneously optimised task predictor,such as a classification or segmentation neural network. In this work, wedevelop transfer learning or adaptation strategies to increase the adaptabilityof both the IQA agent and the task predictor so that they are less dependent onhigh-quality, expert-labelled training data. The proposed transfer learningstrategy re-formulates the original RL problem for task amenability in ameta-reinforcement learning (meta-RL) framework. The resulting algorithmfacilitates efficient adaptation of the agent to different definitions of imagequality, each with its own Markov decision process environment includingdifferent images, labels and an adaptable task predictor. Our work demonstratesthat the IQA agents pre-trained on non-expert task labels can be adapted topredict task amenability as defined by expert task labels, using only a smallset of expert labels. Using 6644 clinical ultrasound images from 249 prostatecancer patients, our results for image classification and segmentation tasksshow that the proposed IQA method can be adapted using data with as few asrespective 19.7% and 29.6% expert-reviewed consensus labels and still achievecomparable IQA and task performance, which would otherwise require a trainingdataset with 100% expert labels.

Adaptable image quality assessment using meta-reinforcement learning of task amenability