Quantitative Susceptibility Mapping in Alzheimer’s Disease using Joint background-field removal and segmentation-Enhanced Dipole Inversion
Jakob Meineke1, Fabian Wenzel1, Iain D. Wilkinson2, and Ulrich Katscher1

1Philips Research Europe, Hamburg, Germany, 2University of Sheffield, Sheffield, United Kingdom


Quantitative Susceptibility Mapping (QSM) is used to study the deep gray-matter nuclei of patients with Alzheimer’s Disease (AD) and healthy control subjects. QSM is performed using “Joint background-field removal and segmentation-Enhanced Dipole Inversion” (JEDI), which leverages the information from automated model-based segmentation and allows the compact single-step formulation of the ill-posed inversion problem of QSM. The tissue magnetic susceptibility shows a trend for increase in the amygdala and the putamen of AD patients as compared to healthy control subjects, in agreement with previous studies.


To assess the potential of Quantitative Susceptibility Mapping (QSM) with Joint background-field removal and segmentation-Enhanced Dipole Inversion (JEDI) for the fully automated measurement of the magnetic susceptibility in subcortical gray-matter nuclei of patients with Alzheimer’s Disease and healthy control subjects.


Disturbed iron regulation and increased iron levels in specific brain regions have long been associated with Alzheimer’s Disease (AD) [1]. The specific role of iron in the framework of AD, its causes and its consequences, are, however, still widely unclear, particularly due to the lack of reliable in vivo measurements. In recent years, MR-based Quantitative Susceptibility Mapping (QSM) has become available to indicate changes in in vivo iron concentration in deep gray matter quantitatively, assuming that magnetic susceptibility of deep gray matter is predominantly determined by ferritin.


12 patients (age: 65$$$\pm$$$11) with AD and 14 healthy controls (HC) (age: 57$$$\pm$$$8) were scanned on a 3T scanner (Ingenia, Philips Healthcare, Best, The Netherlands) using a 32-channel head-coil (University of Sheffield) or 15-channel head-coil (Philips Research Hamburg) after obtaining informed consent from the respective IRBs. The scans are part of an ongoing prospective study within the framework of the EU VPH-DARE@IT project and included a multi-echo gradient-echo sequence for QSM (FOV: (AP, FH, RL) 240×145×210 mm3, true axial orientation, acquisition voxel: 0.6×0.6×2.0 mm3, FA=14°, TE=3.5 ms, ΔTE=4 ms, 7 echoes, TR=31 ms, bipolar readout, BW=275.9 Hz/vx, SENSE (P/S) 1.8x1.2) and a T1-weighted magnetization-prepared TFE sequence which is used for model-based segmentation (FOV: (AP, FH, RL) 240×240×170 mm3, acq voxel 0.94×0.94×1.0 mm3, FA=8°, TR=8 ms, TFE factor=222, inversion delay 1000 ms, BW=191.5 Hz/vx, SENSE (P/S) 1.0/2.2). To solve the ill-posed inverse problem of QSM, the algorithm dubbed “Joint background-field removal and segmentation-Enhanced Dipole Inversion” (JEDI) [2] was chosen, which uses a priori knowledge from automated anatomical segmentation within a single step formulation of the inverse field-to-source problem. In previous studies, JEDI demonstrated improved accuracy and less long-range modulations in numerical simulations and in vivo volunteer data. Reconstruction parameters were identical for all subjects. Within the segmented regions, the median of the susceptibility was computed, referencing to the median of the susceptibility in the corpus callosum.


Figure 1 shows the median susceptibility for segmented brain regions averaged within the AD and the HC groups. Figure 2 shows the difference of the average median susceptibility between the AD and the HC group. The largest increase in susceptibility between the group of AD patients and the HC group are found in the amygdala and the putamen. The differences between groups are about half the standard deviation found within groups. Small differences, below 0.005 ppm, both positive and negative, are found in the caudate nucleus, the thalamus, the hippocampus and the globus pallidus.


The results presented here are in agreement with previous work, which has used alternative reconstruction methods for QSM, and showed elevated susceptibility values for AD patients in the putamen [3] along with indications for an increase in the amygdala. The variability of median susceptibility values within the subject groups is unlikely to be caused by the reconstruction method, and rather reflects subject-specific biological variations. The reproducibility of the QSM reconstruction method used here has been assessed by repeatedly scanning the same volunteer over several months, finding a standard deviation of the median below 0.01 ppm for all brain regions. The effect of age on the magnetic susceptibility depends on the brain region considered [4, 5] and might play a role for the putamen. In a volunteer study (N=19, ages between 32 and 60) performed in connection with the described study, no statistically significant effect of age alone was observed. Larger cohort populations are needed to ascertain whether the observed trends reach statistical significance. The fully automated reconstruction and region-specific evaluation is compatible with a clinical setting.


This project has received funding from European Union’s Seventh Framework Programme, grant no. 601055


1. Yunlong Tao et al., “Perturbed Iron Distribution in Alzheimer’s Disease Serum, Cerebrospinal Fluid, and Selected Brain Regions: A Systematic Review and Meta-Analysis,” Journal of Alzheimer’s Disease 42 (2014): 679

2. Jakob Meineke et al., “Quantitative Susceptibility Mapping Using Segmentation-Enabled Dipole Inversion,” ISMRM 2015, Toronto, Canada, 3321.

3. Julio Acosta-Cabronero et al., “In Vivo Quantitative Susceptibility Mapping (QSM) in Alzheimer’s Disease,” PLoS ONE 8, (2013): e81093

4. Hallgren and Sourander, “The Effect of Age on the Non-Haemin Iron in the Human Brain,” Journal of Neurochemistry 3, (1958): 41

5. Wei Li et al., “Differential Developmental Trajectories of Magnetic Susceptibility in Human Brain Gray and White Matter over the Lifespan,” Human Brain Mapping 35, (2014): 2698


Fig. 1: Median magnetic susceptibility within segmented gray-matter regions, averaged over different subjects within AD and HC groups. Errors bars reflect the standard deviation of the median values.

Fig. 2: Difference of the average susceptibility between AD and HC groups. Positive values reflect larger median susceptibility in the AD group.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)