Hendrikus Joseph Alphons Crooijmans1,
Philippe C. Cattin2, Oliver M. Weber1,3, Klaus
Scheffler1
1Division of Radiological Physics,
Department of Medical Radiology, University of Basel Hospital, Basel,
Switzerland; 2Medical Image Analysis Center (MIAC), University of
Basel, Basel, Switzerland; 3Siemens Healthcare, Zrich,
Switzerland
The
interest point detector and region descriptor SURF finds features in every
image/phase of a series of CINE-bSSFP images. By the help of user input, the
best feature is selected in a Matlab program and corresponding features are
automatically detected. If no feature is found within preset limits, it is
created by linear interpolation. User indicated points of interest are
described as a function of the position and diameter of the feature (all in
the first image/phase), and can therefore be repositioned based on the
corresponding features in the succeeding images/phases. This method provides
reliable fast online valve position detection.