Location: MIC-Arena, Maria-Goeppert-Str. 3, 23562 Lübeck
Title: Topics in Medical Image Registration: Modeling, Accuracy, and Optimization
Abstract: Deformable Image Registration (DIR) methods determine a spatial transformation that best relates the information depicted within a pair or sequence of images with respect to an assumed motion and/or intensity model. Though essential to a host of medical applications, the utility of DIR based analysis is dependent on the spatial accuracy of the recovered transformation. As such, DIR algorithm, validation, and application development are all active areas of research. We recently derived an l1-optimization based DIR method guaranteed to produce globally optimal solutions to a class of parameterized DIR problems. The high spatial accuracy achieved by the method allows for the inference of patient-specific ventilation images from thoracic computed tomography. In this talk, the components required for this type of DIR-based image analysis, including the DIR algorithm, validation framework, and physical modeling, will be discussed.
