A multiresolution adaptive approach for respiratory motion modeling

This paper presents a real-time multiresolution approach for respiratory motion modeling. The proposed methodology utilizes a hybrid approach that combines well-known linear identification methods with a novel global-local orthogonal mapping (GLO-MAP) network. Further, adaptation laws are derived using the recent advances in adaptive control to adapt for respiration models in real-time from imagery data. Finally, the effectiveness of the proposed technique is illustrated by testing them on patient´s chest motion data captured by imagery data of external markers.