eduzhai > Applied Sciences > Engineering >

An Uncertainty-aware Transfer Learning-based Framework for Covid-19 Diagnosis

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 9 pages

Abstract: The early and reliable detection of COVID-19 infected patients is essentialto prevent and limit its outbreak. The PCR tests for COVID-19 detection are notavailable in many countries and also there are genuine concerns about theirreliability and performance. Motivated by these shortcomings, this paperproposes a deep uncertainty-aware transfer learning framework for COVID-19detection using medical images. Four popular convolutional neural networks(CNNs) including VGG16, ResNet50, DenseNet121, and InceptionResNetV2 are firstapplied to extract deep features from chest X-ray and computed tomography (CT)images. Extracted features are then processed by different machine learning andstatistical modelling techniques to identify COVID-19 cases. We also calculateand report the epistemic uncertainty of classification results to identifyregions where the trained models are not confident about their decisions (outof distribution problem). Comprehensive simulation results for X-ray and CTimage datasets indicate that linear support vector machine and neural networkmodels achieve the best results as measured by accuracy, sensitivity,specificity, and AUC. Also it is found that predictive uncertainty estimatesare much higher for CT images compared to X-ray images.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×