eduzhai > Applied Sciences > Engineering >

Image quality assessment for closed-loop computer-assisted lung ultrasound

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 7 pages

Abstract: We describe a novel, two-stage computer assistance system for lung anomalydetection using ultrasound imaging in the intensive care setting to improveoperator performance and patient stratification during coronavirus pandemics.The proposed system consists of two deep-learning-based models: a qualityassessment module that automates predictions of image quality, and a diagnosisassistance module that determines the likelihood-oh-anomaly in ultrasoundimages of sufficient quality. Our two-stage strategy uses a novelty detectionalgorithm to address the lack of control cases available for training thequality assessment classifier. The diagnosis assistance module can then betrained with data that are deemed of sufficient quality, guaranteed by theclosed-loop feedback mechanism from the quality assessment module. Using morethan 25000 ultrasound images from 37 COVID-19-positive patients scanned at twohospitals, plus 12 control cases, this study demonstrates the feasibility ofusing the proposed machine learning approach. We report an accuracy of 86 whenclassifying between sufficient and insufficient quality images by the qualityassessment module. For data of sufficient quality - as determined by thequality assessment module - the mean classification accuracy, sensitivity, andspecificity in detecting COVID-19-positive cases were 0.95, 0.91, and 0.97,respectively, across five holdout test data sets unseen during the training ofany networks within the proposed system. Overall, the integration of the twomodules yields accurate, fast, and practical acquisition guidance anddiagnostic assistance for patients with suspected respiratory conditions atpoint-of-care.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×