Autoadaptive motion modelling

Respiratory motion is a complicating factor for many applications in medical imaging. Respiration is an approximately periodic, but very complex motion that may undergo significant changes within the duration of a treatment or imaging session. Motion models are a possible solution to the problem of respiratory motion. However, in the current state-of-the-art, the model is formed preprocedure and may lose validity during the procedure. We propose a novel autoadaptive motion model which can automatically adapt to changing breathing patterns and thus maintain its validity. We quantitatively evaluated the method on synthetic data generated from MR images acquired from 4 healthy volunteers and found that motion estimation errors after a change in breathing pattern were significantly reduced using the proposed method. Furthermore, we demonstrated the method on real MR data acquired from one healthy volunteer.