eduzhai > Applied Sciences > Engineering >

Lesion Mask-based Simultaneous Synthesis of Anatomic and MolecularMR Images using a GAN

  • Save

... pages left unread,continue reading

Document pages: 10 pages

Abstract: Data-driven automatic approaches have demonstrated their great potential inresolving various clinical diagnostic dilemmas for patients with malignantgliomas in neuro-oncology with the help of conventional and advanced molecularMR images. However, the lack of sufficient annotated MRI data has vastlyimpeded the development of such automatic methods. Conventional dataaugmentation approaches, including flipping, scaling, rotation, and distortionare not capable of generating data with diverse image content. In this paper,we propose a method, called synthesis of anatomic and molecular MR imagesnetwork (SAMR), which can simultaneously synthesize data from arbitrarymanipulated lesion information on multiple anatomic and molecular MRIsequences, including T1-weighted (T1w), gadolinium enhanced T1w (Gd-T1w),T2-weighted (T2w), fluid-attenuated inversion recovery (FLAIR), and amideproton transfer-weighted (APTw). The proposed framework consists of astretch-out up-sampling module, a brain atlas encoder, a segmentationconsistency module, and multi-scale label-wise discriminators. Extensiveexperiments on real clinical data demonstrate that the proposed model canperform significantly better than the state-of-the-art synthesis methods.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×