[en] Breathing motion is a significant source of error in radiotherapy treatment of malignant thoracic and abdominal tumours: Accounting for breathing motion by increasing treatment margins is not desirable as it substantially increases the volume of irradiated healthy tissues. So attempts to explicitly account for breathing motion are underway (like 4D radiotherapy) but suffer from little existing knowledge regarding the spatial-temporal behaviour of anatomical and pathological structures involved. The objective of this thesis is to analyse breathing motion of the left and right lung, the diaphragm, the chest wall, and lung tumours. To accomplish such studies the authors use spatial-temporal CT image sequences acquired during free breathing. Patients examined are lung tumour patients.A multi slice CT scanner is used to acquire the spatial-temporal CT image sequences. But, such scanners can only simultaneously scan a limited region of the body and a limited anatomical segment respectively. So we reconstruct 3D CT data sets at different points of the breathing cycle, where each 3D data set covers all anatomical segments available (in this case: whole thorax and parts of the upper abdomen). The reconstructed 4D CT data sets provide a basis for representing the spatial-temporal behaviour of the structures to be considered. A voxel based modelling approach is used. The structures' spatial extension gets represented by a 3D binary image sequence, where each of the data sets corresponds to a 3D CT image out of the 4D CT data set. In addition the motion occurring between these points in time is presented by voxel based displacement fields. The fields are calculated using non-linear registration. As registration basis (fixed image and moving image) either the binary data sets or the 3D CT data sets can be chosen. In the present cases the binary data sets show better results