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

Objective: Medical image datasets with pixel-level labels tend to have alimited number of organ or tissue label classes annotated, even when the imageshave wide anatomical coverage. With supervised learning, multiple classifiersare usually needed given these partially annotated datasets. In this work, wepropose a set of strategies to train one single classifier in segmenting alllabel classes that are heterogeneously annotated across multiple datasetswithout moving into semi-supervised learning. Methods: Masks were first createdfrom each label image through a process we termed presence masking. Threepresence masking modes were evaluated, differing mainly in weightage assignedto the annotated and unannotated classes. These masks were then applied to theloss function during training to remove the influence of unannotated classes.Results: Evaluation against publicly available CT datasets shows that presencemasking is a viable method for training class-generic classifiers. Ourclass-generic classifier can perform as well as multiple class-specificclassifiers combined, while the training duration is similar to that requiredfor one class-specific classifier. Furthermore, the class-generic classifiercan outperform the class-specific classifiers when trained on smaller datasets.Finally, consistent results are observed from evaluations against human thighand calf MRI datasets collected in-house. Conclusion: The evaluation outcomesshow that presence masking is capable of significantly improving both trainingand inference efficiency across imaging modalities and anatomical regions.Improved performance may even be observed on small datasets. Significance:Presence masking strategies can reduce the computational resources and costsinvolved in manual medical image annotations. All codes are publicly availableat https://github.com/wong-ck/DeepSegment.

Training CNN Classifiers for Semantic Segmentation using Partially Annotated Images: with Application on Human Thigh and Calf MRI