1 Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark2 Department of Informatics and Mathematical Modeling, Technical University of Denmark3 Biomedical Engineering, Department of Electrical Engineering, Technical University of Denmark4 Department of Electrical Engineering, Technical University of Denmark5 Copenhagen Center for Health Technology, Center, Technical University of Denmark
Motion correction in MR spectroscopy Downscatter correction in SPECT
The quality of medical scanner data is often compromised by several mechanisms. This can be caused by both the subject to be measured and the scanning principles themselves. In this PhD project the problem of subject motion was addressed for Single Voxel MR Spectroscopy in a cohort study of preterm infants. In Iodine-123 SPECT the problem of downscatter was addressed. This thesis is based on two papers. Paper I deals with the problem of motion in Single Voxel Spectroscopy. Two novel methods for the identification of outliers in the set of repeated measurements were implemented and compared to the known mean and median filtering. The data comes from non-anesthetized preterm infants, where motion during scanning is a common problem. Both the novel outlier identification and the independent component analysis (ICA) perform satisfactory and better than the common mean and median filtering. ICA performed best in the sense that it recovered most of the lost peak height in the spectra. The ICA motion correction algorithm described in paper I and in this thesis was applied to a quantitative analysis of the Single Voxel Spectroscopy data from the cohort study of preterm infants. This analysis revealed that differences between term and preterm infants are not to be found in the concentrations of Lactate (caused by inflammation or hypoxia-ischemia) and/or NAA (caused by hypoxia-ischemia) as hypothesized before the cohort study. Instead choline levels were decreased in the preterm infants, which might indicate a detrimental effect of the extra-uterine environment on brain development. Paper II describes a method to correct for downscatter in low count Iodine-123 SPECT with a broad energy window above the normal imaging window. Both spatial dependency and weight factors were measured. As expected, the implicitly assumed weight factor of one for energy windows with equal width is slightly too low, due the presence of a backscatter peak in the energy spectrum coming from high-energy photons. The effect on the contrast was tested in 10 subjects and revealed a 20% increase in the specific binding ratio of the striatum due to downscatter correction. This makes the difference between healthy subjects and patients more profound. Downscatter in Iodine-123 SPECT is not the only deteriorating mechanism. Normal scatter compromises the images quality as well. Since scatter correction of SPECT-images also can be performed by the subtraction of an energy window, a method was developed to perform scatter and downscatter correction simultaneously. A phantom study has been performed, where the in paper II described downscatter correction was extended with scatter correction. This new combined correction was compared to the known Triple Energy Window (TEW) correction method. Results were satisfying and indicate that TEW is more correct from the physics point of view, while the in paper II described method extended with scatter correction gives reasonable results, but is far less noise sensitive than TEW.