Seminar
Automatic Partial Facial Nerve Canal Segmentation Using A Multidimensional Approach
Date: Thursday 14th May 2009
Commencing: 12.00 pm
Venue: ICT Theatre 2, Ground Floor, Room 20,
ICT Building, 111 Barry Street, Carlton
Presented by: Mr Mauro Maiorca
Software Engineer/ Researcher, MUVES project , Univeristy of Melbourne
Traumatic injury to the facial nerve is a serious complication of otologic surgery. To avoid this complication, it is advantageous for surgeons to know the location of the facial nerve prior to surgery.
The location of the facial nerve is estimated by identifying the facial nerve canal in patient CT data. Performing this task manually takes hours and is error prone. In this paper we present an unsupervised algorithm for automatically segmenting the facial nerve canal in CT data using intensity, size and shape. To test our algorithm we statistically compared its predictions with ground truth for 41 CT scans; most cases presented serious pathologies or abnormalities. Its resulting mean entropy correlation coefficient is 0.5 (SD 0.15), thus our approach may suit clinical application.
*Venue:
The University of Melbourne,
Theatre 2, ICTBuilding
111 Barry Street, Carlton
(Building 105, University Map)