eduzhai > Applied Sciences > Engineering >

Deep Placental Vessel Segmentation for Fetoscopic Mosaicking

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 11 pages

Abstract: During fetoscopic laser photocoagulation, a treatment for twin-to-twintransfusion syndrome (TTTS), the clinician first identifies abnormal placentalvascular connections and laser ablates them to regulate blood flow in bothfetuses. The procedure is challenging due to the mobility of the environment,poor visibility in amniotic fluid, occasional bleeding, and limitations in thefetoscopic field-of-view and image quality. Ideally, anastomotic placentalvessels would be automatically identified, segmented and registered to createexpanded vessel maps to guide laser ablation, however, such methods have yet tobe clinically adopted. We propose a solution utilising the U-Net architecturefor performing placental vessel segmentation in fetoscopic videos. The obtainedvessel probability maps provide sufficient cues for mosaicking alignment byregistering consecutive vessel maps using the direct intensity-based technique.Experiments on 6 different in vivo fetoscopic videos demonstrate that thevessel intensity-based registration outperformed image intensity-basedregistration approaches showing better robustness in qualitative andquantitative comparison. We additionally reduce drift accumulation tonegligible even for sequences with up to 400 frames and we incorporate a schemefor quantifying drift error in the absence of the ground-truth. Our paperprovides a benchmark for fetoscopy placental vessel segmentation andregistration by contributing the first in vivo vessel segmentation andfetoscopic videos dataset.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...