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

As deep learning is widely used in the radiology field, the explainability ofsuch models is increasingly becoming essential to gain clinicians' trust whenusing the models for diagnosis. In this research, three experiment sets wereconducted with a U-Net architecture to improve the classification performancewhile enhancing the heatmaps corresponding to the model's focus throughincorporating heatmap generators during training. All of the experiments usedthe dataset that contained chest radiographs, associated labels from one of thethree conditions ("normal", "congestive heart failure (CHF)", and "pneumonia"),and numerical information regarding a radiologist's eye-gaze coordinates on theimages. The paper (A. Karargyris and Moradi, 2021) that introduced this datasetdeveloped a U-Net model, which was treated as the baseline model for thisresearch, to show how the eye-gaze data can be used in multi-modal training forexplainability improvement. To compare the classification performances, the 95%confidence intervals (CI) of the area under the receiver operatingcharacteristic curve (AUC) were measured. The best method achieved an AUC of0.913 (CI: 0.860-0.966). The greatest improvements were for the "pneumonia" and"CHF" classes, which the baseline model struggled most to classify, resultingin AUCs of 0.859 (CI: 0.732-0.957) and 0.962 (CI: 0.933-0.989), respectively.The proposed method's decoder was also able to produce probability masks thathighlight the determining image parts in model classifications, similarly asthe radiologist's eye-gaze data. Hence, this work showed that incorporatingheatmap generators and eye-gaze information into training can simultaneouslyimprove disease classification and provide explainable visuals that align wellwith how the radiologist viewed the chest radiographs when making diagnosis.

Improving Disease Classification Performance and Explainability of Deep Learning Models in Radiology with Heatmap Generators