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

Locating diseases in chest X-ray images with few careful annotations saveslarge human effort. Recent works approached this task with innovativeweakly-supervised algorithms such as multi-instance learning (MIL) and classactivation maps (CAM), however, these methods often yield inaccurate orincomplete regions. One of the reasons is the neglection of the pathologicalimplications hidden in the relationship across anatomical regions within eachimage and the relationship across images. In this paper, we argue that thecross-region and cross-image relationship, as contextual and compensatinginformation, is vital to obtain more consistent and integral regions. To modelthe relationship, we propose the Graph Regularized Embedding Network (GREN),which leverages the intra-image and inter-image information to locate diseaseson chest X-ray images. GREN uses a pre-trained U-Net to segment the lung lobes,and then models the intra-image relationship between the lung lobes using anintra-image graph to compare different regions. Meanwhile, the relationshipbetween in-batch images is modeled by an inter-image graph to compare multipleimages. This process mimics the training and decision-making process of aradiologist: comparing multiple regions and images for diagnosis. In order forthe deep embedding layers of the neural network to retain structuralinformation (important in the localization task), we use the Hash coding andHamming distance to compute the graphs, which are used as regularizers tofacilitate training. By means of this, our approach achieves thestate-of-the-art result on NIH chest X-ray dataset for weakly-superviseddisease localization. Our codes are accessible online(https://github.com/qibaolian/GREN).

GREN: Graph-Regularized Embedding Network for Weakly-Supervised Disease Localization in X-ray Images