eduzhai > Applied Sciences > Engineering >

Semi-Supervised Learning for Fetal Brain MRI Quality Assessment with ROI consistency

  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: Fetal brain MRI is useful for diagnosing brain abnormalities but ischallenged by fetal motion. The current protocol for T2-weighted fetal brainMRI is not robust to motion so image volumes are degraded by inter- and intra-slice motion artifacts. Besides, manual annotation for fetal MR image qualityassessment are usually time-consuming. Therefore, in this work, asemi-supervised deep learning method that detects slices with artifacts duringthe brain volume scan is proposed. Our method is based on the mean teachermodel, where we not only enforce consistency between student and teacher modelson the whole image, but also adopt an ROI consistency loss to guide the networkto focus on the brain region. The proposed method is evaluated on a fetal brainMR dataset with 11,223 labeled images and more than 200,000 unlabeled images.Results show that compared with supervised learning, the proposed method canimprove model accuracy by about 6 and outperform other state-of-the-artsemi-supervised learning methods. The proposed method is also implemented andevaluated on an MR scanner, which demonstrates the feasibility of online imagequality assessment and image reacquisition during fetal MR scans.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×