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

The COVID-19 pandemic has inspired unprecedented data collection and computervision modelling efforts worldwide, focusing on diagnosis and stratification ofCOVID-19 from medical images. Despite this large-scale research effort, thesemodels have found limited practical application due in part to unprovengeneralization of these models beyond their source study. This studyinvestigates the generalizability of key published models using the publiclyavailable COVID-19 Computed Tomography data through cross dataset validation.We then assess the predictive ability of these models for COVID-19 severityusing an independent new dataset that is stratified for COVID-19 lunginvolvement. Each inter-dataset study is performed using histogramequalization, and contrast limited adaptive histogram equalization with andwithout a learning Gabor filter. The study shows high variability in thegeneralization of models trained on these datasets due to varied sample imageprovenances and acquisition processes amongst other factors. We show that undercertain conditions, an internally consistent dataset can generalize well to anexternal dataset despite structural differences between these datasets with f1scores up to 86%. Our best performing model shows high predictive accuracy forlung involvement score for an independent dataset for which expertly labelledlung involvement stratification is available. Creating an ensemble of our bestmodel for disease positive prediction with our best model for disease negativeprediction using a min-max function resulted in a superior model for lunginvolvement prediction with average predictive accuracy of 75% for zero lunginvolvement and 96% for 75-100% lung involvement with almost linearrelationship between these stratifications.

Systematic investigation into generalization of COVID-19 CT deep learning models with Gabor ensemble for lung involvement scoring