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

Yes, it can. Data augmentation is perhaps the oldest preprocessing step incomputer vision literature. Almost every computer vision model trained onimaging data uses some form of augmentation. In this paper, we use theinter-vertebral disk segmentation task alongside a deep residual U-Net as thelearning model, to explore the effectiveness of augmentation. In the extreme,we observed that a model trained on patches extracted from just one scan, witheach patch augmented 50 times; achieved a Dice score of 0.73 in a validationset of 40 cases. Qualitative evaluation indicated a clinically usablesegmentation algorithm, which appropriately segments regions of interest,alongside limited false positive specks. When the initial patches are extractedfrom nine scans the average Dice coefficient jumps to 0.86 and most of thefalse positives disappear. While this still falls short of state-of-the-artdeep learning based segmentation of discs reported in literature, qualitativeexamination reveals that it does yield segmentation, which can be amended byexpert clinicians with minimal effort to generate additional data for trainingimproved deep models. Extreme augmentation of training data, should thus beconstrued as a strategy for training deep learning based algorithms, when verylittle manually annotated data is available to work with. Models trained withextreme augmentation can then be used to accelerate the generation of manuallylabelled data. Hence, we show that extreme augmentation can be a valuable toolin addressing scaling up small imaging data sets to address medical imagesegmentation tasks.

Extreme Augmentation : Can deep learning based medical image segmentation be trained using a single manually delineated scan?