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/

Zendy Plus

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

Zendy Tools

Zendy Open

Cervical cancer is the fourth most common cancer in women worldwide. Theavailability of a robust automated cervical image classification system canaugment the clinical care provider's limitation in traditional visualinspection with acetic acid (VIA). However, there are a wide variety ofcervical inspection objectives which impact the labeling criteria forcriteria-specific prediction model development. Moreover, due to the lack ofconfirmatory test results and inter-rater labeling variation, many images areleft unlabeled.  Motivated by these challenges, we propose a self-supervised learning (SSL)based approach to produce a pre-trained cervix model from unlabeled cervicalimages. The developed model is further fine-tuned to produce criteria-specificclassification models with the available labeled images. We demonstrate theeffectiveness of the proposed approach using two cervical image datasets. Bothdatasets are partially labeled and labeling criteria are different. Theexperimental results show that the SSL-based initialization improvesclassification performance (Accuracy: 2.5% min) and the inclusion of imagesfrom both datasets during SSL further improves the performance (Accuracy: 1.5%min). Further, considering data-sharing restrictions, we experimented with theeffectiveness of Federated SSL and find that it can improve performance overthe SSL model developed with just its images. This justifies the importance ofSSL-based cervix model development. We believe that the present research showsa novel direction in developing criteria-specific custom deep models forcervical image classification by combining images from different sourcesunlabeled and/or labeled with varying criteria, and addressing image accessrestrictions.

Deep Cervix Model Development from Heterogeneous and Partially Labeled Image Datasets