Segmentation and mosaicking of intraoperative fetoscopic video for fetal surgeries

Author: Carmen García Nogueiras

Virtual Room: https://eu.bbcollab.com/guest/96240734673541a99836ea2372e39c9b

Date & time: 22/09/2021 – 9:30 h

Session name: Medical Imaging

Supervisor: Mario Ceresa

Abstract:

Twin-to-twin transfusion syndrome (TTTS) is a pregnancy condition affecting identical twins. In these cases, the blood flow among the twins is uneven. The most common treatment consists in photocoagulating the abnormal connections (anastomoses). During this intervention, a fetoscopic camera inspects the placenta’s surface. This device has a limited field of view (FoV) and manoeuvrability. In addition, the quality of the obtained images is poor due to the lower resolution of the cameras and the amniotic fluid turbidity. To tackle this, we propose a mosaicking algorithm supported by a deep semantic segmentation of the placenta. Finally, we present qualitative and quantitative results and discuss their suitability within the frame of ASTRA objectives.

Committee:

– President: Maria Vanrell Martorell(UAB)
– Secretary: Jorge Bernal(UAB)
– Vocal: Fernando Vilariño(UAB)