Treatment of children
with congenital craniofacial malformations or individuals that acquire face or
head trauma often includes the use of 3D imaging. In order to quantify change
over time due to treatment (including surgery) and/or growth, it is necessary
to carry out a comparison between two or more 3D images. This requires
determination of point correspondence from one image (e.g. obtained before
surgery) to another image (e.g. obtained after surgery).
This project will
explore the use of image registration algorithms in order to determine
so-called detailed point correspondence between a set of MR images of the same
individual obtained at different times. Based on non-rigid image registration
algorithms that quantify how to deform one image to best math another image, a
deformation field may be studied in order to quantify and analyze the amount,
localization and magnitude of change that has taken place in the face and jaws.
Interpretation of the result depends on the use of a valid common reference
system for all the involved images.
It is proposed to use the inner ear
(semi-circular canals and/or cochlea; Figure 1) as fiducials for the
registration, and a focus of this project will be to evaluate the pros and cons
of the use of this reference system. A particular advantage is the known
stability of the inner ear as it is fully formed before birth. The project
should carry out an error analysis in order to determine precision of the
measurement, and provide guidelines describing the ability of this method to be
used in practice for treatment evaluation purposes.
The project has access to a
unique longitudinal data material on more than 30 normal children and
adolescents where MR scans have been obtained 10 times with a time interval of
half a year. The project is a collaboration with 3D Lab and Danish Research Center for
Magnetic Resonance at Hvidovre Hospital.
I samarbejde med3D Lab, DRCMR
It is an advantage with earlier experience / knowledge / courses in image analysis, but not an absolute requirement. Some programming experience (e.g. in Matlab or C++) is an advantage.