Entwicklung schneller und robuster Messtechniken zur verbesserten radiologischen Diagnostik durch 3D Parameterquantifizierung für die Magnet-Resonanz-Tomographie (MRT)
Kognitive und systemische Humanneurowissenschaften
Zusammenfassung der Projektergebnisse
Since the acquisition of the first human magnetic resonance (MR) images in the early 1980s, MRI has played an ever increasing role in medical diagnosis largely due to its non-invasive nature and its abundance in contrast possibilities and richness in information content. In current clinical practice, several individual MRI sequences are applied one after another in order to generate multiple images with different contrasts in multiple orientations. However, this procedure can be rather time consuming, prone to misregistration and usually delivers anisotropic spatial resolution. In addition, these images often have suboptimal contrast and, unlike in computed tomography (CT), do not contain quantitative information at all. The overall objective of this project has been the technical development of a fast and robust 3D MRI sequence for the simultaneous quantification of proton-density (PD) and relaxation parameters T1 and T2 with isotropic submillimeter spatial resolution. From these volume data sets, synthetic images with virtually arbitrary contrasts in arbitrary slice orientation can be generated after the fact with the potential of improved pathology detection. Unfortunately, the ultimate goal to provide accurate 3D quantitative relaxation and spin density parameter maps in submillimeter resolution in each dimension within a scan time below 10 min could not fully be reached.. However, we were able to show that accurate, high quality parameter maps can be obtained with 0.9 x 0.9 x 3 mm3 resolution within only 7.5 min scan time. Thus, we are confident that our ultimate goal can be reached in the near future by adding further acceleration of about a factor 2 (e.g. higher parallel imaging factors, partial Fourier, compressed sensing) to the stack of stars IR-sequence. Alternatively, an IR-TrueFisp with a 3D radial or Cone readout could potentially help to achieve this goal. However, we were not able to target this approach within the funding period. The great advantage of IR-TrueFisp, namely to simultaneously provide T1/T2/SD perfectly registered has been the main reason for us to concentrate our developments around this sequence type. However, this benefit comes with the drawback that the T1/T2/SD values derived are very sensitive to B0 inhomogeneities and flip-angle profiles and in practice often result in an over or under estimation of the actual parameters. Thus, we put great effort into the development of several approaches allowing us to successfully correcting for the erroneous parameter maps. Specifically, we derived analytic expressions which allow us to correct the T1/T2 values when the off-resonance in this voxel is known. Thus, a short pre-scan to estimate the off-resonance in each voxel can be carried out and used to correct the acquired parameter maps. Another important development has been the PCA based reconstruction which allowed to significantly improve the accuracy of spin-density maps compared to a KWIC filter based reconstruction. Furthermore, as almost all existing quantitative imaging approaches assume single compartment models, we extended the approach such that multiple components can be detected using the IR-TrueFISP technique. The measured data along an IR-Trufisp readout represent multi-exponential signals which can be analyzed voxel-by-voxel using the inverse Laplace transform. The results can be represented as relaxation time spectra, from which both T1 and T2 can potentially be extracted for each component in the voxel. Finally, an alternative approach for simultaneous quantitative parameter mapping in 3D, namely dynamically phase-cycled (DPC) TrueFISP has been investigated towards the ability for providing robust high-quality parameter maps at high field strengths without being sensitive to off-resonance effects. Preliminary work has been conducted on a phantom in order to investigate the DPC TrueFISP method as potential candidate for a more robust and more timeefficient sequence. In conclusion, although not all aspects could be addressed as written in the proposal, we still believe that the results we obtained show great potential to reach the ultimate goal of fast and robust 3D MR parameter mapping in high spatial resolution and thus may fundamentally change the way clinical MRI exams will be carried out in the future.
Projektbezogene Publikationen (Auswahl)
- Schnelle Simultane Detektion nicht aktiver und schrankengestörter Entmarkungsherde bei Multipler Sklerose (MS) mit einer quantifizierbaren MRT-Multikontrastsequenz (qMRI). Book of abstracts: 50. Jahrestagung der Deutschen Gesellschaft für Neuroradiologie (DGNR), Köln (2015): Abstract 23
Homola G, Blaimer M, Kleesiek J, Breuer F, Bartsch AJ
(Siehe online unter https://dx.doi.org/10.1007/s00062-015-0445-4) - Fast spatially-resolved multi-component T1 and T2 parameter mapping. Book of abstracts of the 34th Annual Scientific Meeting of the ESMRMB, Barcelona, Spain (2017): Abstract 111
Pfister J, Breuer F, Jakob P, Blaimer M
- Simultaneous T1/T2 measurements in combination with PCA-SENSE reconstruction (T1* shuffling) and multicomponent analysis. Proceedings of the 25th Annual Meeting of the ISMRM, Honolulu, HI, USA (2017): Abstract 452
Pfister J, Blaimer M, Jakob P, Breuer F
- Towards 3D single sequence simultaneous T1- and T2-quantification with multicomponent analysis. Book of abstracts of the 34th Annual Scientific Meeting of the ESMRMB, Barcelona, Spain (2017): Abstract 110
Pfister J, Blaimer M, Jakob P, Breuer F
