Inter- and intra-subject variation of abdominal vs. thoracic respiratory motion using kernel density estimation

In nuclear medicine, there is a significant research focus in developing a new approach in monitoring, tracking and compensating respiratory motion during image acquisition. We address this by attempting to model the respiratory cycle pattern and finding a method that describes the configuration of the anterior surface which then correlates with the internal position/configuration of the internal organ as a foundation for motion compensation. This paper presents novel work in parameterizing external respiratory motion using a method based on the variation of abdominal vs. thoracic surface markers to investigate inter- and intra-subject variation. The dominant mode of variation of the Abdominal and Thoracic surfaces during respiration using Principle Component Analysis (PCA) is studied. This demonstrates that pattern of TS vs AS motion appears temporally at a global level stable. Thus although breathing style is consistent within a given subject, we there observe temporal changes in the amplitude and density of the PDF in intra-subject data