FAIR-ASL revisited: improving accuracy of ASL perfusion MRI.

Fabian T. Gutjahr (Würzburg)1, M. Gram (Würzburg)2, T. Kampf (Würzburg)3, P. M. Jakob (Würzburg)1, W. R. Bauer (Würzburg)4, M. Christa (Würzburg)2

1Universität Würzburg Lehrstuhl Experimentelle Physik V Würzburg, Deutschland; 2Universitätsklinikum Würzburg Medizinische Klinik und Poliklinik I Würzburg, Deutschland; 3Universitätsklinikum Würzburg Diagnostische und interventionelle Neuroradiologie Würzburg, Deutschland; 4Universitätsklinikum Würzburg Med. Klinik und Poliklinik I, Klinische Elektrophysiologie, Kardio MRT Würzburg, Deutschland



Arterial spin labeling(ASL) is a magnetic resonance imaging(MRI) technique that enables the contrast agent free measurement of perfusion. In ASL, inflowing arterial blood is manipulated using labeling slices in proximity to the measured slice. However, as the myocardium is supplied in a short loop it is difficult to measure, for cardiac perfusion measurements, the Flow Sensitive Alternating Inversion Recovery (FAIR)-ASL technique has become the predominant technique. In FAIR, perfusion is encoded in the difference between the T1-times estimated after either global (labeling) or slice selective inversion (reference). This has been shown to be a reliable and robust technique [1]. However, FAIR-ASL assumes, that inflowing blood in the reference experiment is not manipulated by the inversion pulse. This assumption is not valid in practice, as blood in the heart is also labeled in the reference experiment (see Figure 1). Therefore FAIR-ASL will underestimate perfusion. This can be especially problematic, as the underestimation is dependent on heart rate and ejection fraction. We propose to replace the reference experiment by measuring the transient magnetization evolution from equilibrium(M0) to steady state(Minf) to reduces the amount of unwanted labeling.


Methods and Materials

Experiments were conducted on a 7T small animal system. The retrospectively triggered inversion recovery snapshot FLASH based perfusion measurement as described in [2] was implemented and adapted to allow the measurement and model based reconstruction of the transient magnetization evolution. All animals were measured using a) a global inversion preparation, b) a slice selective inversion preparation and c) no preparation pulse. Perfusion maps were calculated from measurements a)+b) for FAIR-ASL and a)+c) for the novel method.

C57BL/6 mice underwent either SHAM or MI surgery and were measured via MRI eight weeks after. Sequence parameters: Matrix 128x88, FOV 30x20mm2, imaging/inversion slice-thickness 1.5mm/4.5 mm, TE/TR 1.22ms/3.6 ms, 2794 readouts per inversion, 1760 readouts for the transient, waiting time with a) a slice Selective and b) a global inversion, as well as c) no inversion: 10s.



The novel method - without any improvements for time efficient acquisition - lead to a a measurement time 36 minutes, compared to 20 minutes for the old FAIR-ASL. Direct comparison of perfusion maps acquired with these methods, show an increase of median perfusion by 23±9% with the novel method. (mean±SD; n=5).(see Figure 2).


Discussion and Conclusion

As predicted by the model, first measurements show increased perfusion values. We attribute these changes to the successful reduction of unwanted labeling of inflowing magnetization. These results are especially meaningful if healthy and animals with reduced EF will be compared, as the EF-dependent underestimation will be less pronounced.

By measuring the evolution from equilibrium (M0) to steady state  (Minf) instead of the evolution after Inversion (-M0) to (Minf) the total dynamic range of the reference experiment is reduced (see Figure 3). In the measurements shown here, the corresponding SNR-loss was countered using more averages, resulting in a longer total measurement time, which is a limitation of the novel method. In future studies we will study methods to reduce the SNR penalty of the novel method and assess the dependence of the underestimation on the ejection fraction using animals with reduced EF.

Diese Seite teilen