eduzhai > Applied Sciences > Engineering >

A Data-Efficient Deep Learning Based Smartphone Application For Detection Of Pulmonary Diseases Using Chest X-rays

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 22 pages

Abstract: This paper introduces a paradigm of smartphone application based diseasediagnostics that may completely revolutionise the way healthcare services arebeing provided. Although primarily aimed to assist the problems in renderingthe healthcare services during the coronavirus pandemic, the model can also beextended to identify the exact disease that the patient is caught with from abroad spectrum of pulmonary diseases. The app inputs Chest X-Ray imagescaptured from the mobile camera which is then relayed to the AI architecture ina cloud platform, and diagnoses the disease with state of the art accuracy.Doctors with a smartphone can leverage the application to save the considerabletime that standard COVID-19 tests take for preliminary diagnosis. The scarcityof training data and class imbalance issues were effectively tackled in ourapproach by the use of Data Augmentation Generative Adversarial Network (DAGAN)and model architecture based as a Convolutional Siamese Network with attentionmechanism. The backend model was tested for robustness us-ing publiclyavailable datasets under two different classificationscenarios(Binary Multiclass) with minimal and noisy data. The model achievedpinnacle testing accuracy of 99.30 and 98.40 on the two respective scenarios,making it completely reliable for its users. On top of that a semi-livetraining scenario was introduced, which helps improve the app performance overtime as data accumulates. Overall, the problems of generalisability of complexmodels and data inefficiency is tackled through the model architecture. The appbased setting with semi live training helps in ease of access to reliablehealthcare in the society, as well as help ineffective research of rarediseases in a minimal data setting.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×