Respiratory Motion Estimation: Tests and Comparison of Different Sensors

Physics IDP together with the Nuclear Medicine Department, Klinikum Rechts der Isar der TU München. Though not mentioned in the title, the project had a strong affiliation to PET imaging. A rather detailed overview of the physics of PET is given in the introduction of the report. It can be suggested as reading for other people starting to work in the context of PET.

Summary

The principles of respiratory gating.

Respiratory motion has negative impact on some medical imaging techniques, in particular on Positron Emission Tomography (PET). This is due to the physical limitations of PET, which lead to long acquisition times of several minutes. For high quality PET images of the thorax area, respiratory motion needs to be measured and corrected. This can be done by respiratory gating, a method dividing the respiratory cycle into phases of similar thorax expansion, called gates. Separate images are then reconstructed from the PET data acquired during each such gate.

Respiration gates can be calculated automatically from a respiratory signal, which has to be measured by an external sensor. The purpose of this work was to evaluate four respiration sensors based on different physical principles: an elastic belt measuring thorax circumference, a spirometer measuring air-flow from the lungs, a thermometer measuring the temperature of respired air, and a stereo infrared camera system tracking thorax motion. Focus was put on the camera system, which delivers raw motion data of markers attached to the thorax, and can therefore be used to estimate motion more generally. A feasible setup of the markers and the camera is described, and several processing steps applied to the raw tracking data are explained. This includes interpolation of temporarily occluded markers and calibration of marker locations with respect to the PET scanner. Methods to digest respiratory signals from the multi-dimensional tracking data are also described.

Clinical setup of the camera system for respiration measurements.

The respiratory signals acquired with all the sensors were compared both visually and statistically. All sensors were able to track both regular breathing, and breathing artifacts like breath-holds or speaking. Though based on different physical effects, all respiration curves were basically suited for respiratory gating. However, only the elastic belt and the camera system used in this work, were reliable enough for clinical use, and were tested in a preliminary study with cardiac patients undergoing PET. Gated PET images, reconstructed using the respective respiratory signals, showed respiration-induced motion of the heart clearly, though its magnitude was only in the region of half a centimeter.

The camera system also allowed quantification of respiratory motion on the patient's surface, which turned out have a similar magnitude. Measurements of translative body motion not related to respiration showed that it was usually even smaller, and therefore has little impact on PET imaging. However, more general measurements of body motion are discussed, since non-rigid motion could be much more significant. Only if its characteristics are known, can such body motion be corrected in the future.

In conclusion, respiratory gating can improve the quality of PET imaging, no matter what sensor it is based on. However, the camera system is too cumbersome for respiratory gating alone, and other respiration sensors, like the elastic belt are more advisable for clinical use. Yet, the camera system should be used to study motion of a larger number of patients, to estimate its impact on PET imaging.

Files

Here is the report of the IDP Respiratory Motion Estimation - Tests and Comparison of Different Sensors as PDF (4.8 MB). The TeX sources can be found in the SVN repository.

The Presentation slides for this IDP were created with OpenOffice and are available in OpenDocument Format: Respiratory Motion Estimation - Tests and Comparison of Different Sensors.odp. Since certain popular comercial products don't support this standard yet, here is the same version in MS Power Point format: Respiratory Motion Estimation - Tests and Comparison of Different Sensors.ppt. (However, the latter one is not supposed to look good. Try to use the first one, if possible!)

The two small software packages created in this IDP are written in IDL 6.0 under Linux, but should work under Windows, too. ARTCamera.zip contains a software library for processing raw ART camera data in the context of respiratory motion estimation. SignalAnalysis.zip contains a GUI application for extracting respiratory signals from ART camera data, and from other respiration sensors. It also supports comparison of different respiratory signals.