A cohort of 45 AD patients (average age=71.8, SD=10.8) and 19 DLB patients (average age=70.5, SD=6.4) underwent 3T-MRI scans using both the six-echo ISMRM and ADNI4 MEGRE protocols. To evaluate the impact of the number of echoes, the standard five-echo ADNI4 acquisition was modified to include an additional sixth echo at the end of the sequence. STI Suite, implemented in Matlab, was used to generate QSM maps in the SNc.

Spearman correlation:

The correlation of ISMRM and ADNI4-5echo is shown on the left, and the ISMRM and ADNI4-6echo is shown in the middle. Adding a sixth echo to the ADNI4 protocol did not meaningfully affect its correlation with the ISMRM protocol. On the right, the correlation of the ADNI4-5echo and 6echo protocols is shown. The very high correlation suggests that the additional echo has minimal impact on SNc QSM values.

ROC Analysis: 

ROC analysis showed the highest area under the curve (AUC) with the five-echo ADNI4 (AUC=0.74), followed closely by the six-echo ADNI4 protocol (AUC=0.73), and finally the ISMRM protocol (AUC=0.67) for distinguishing AD and DLB. DeLong test comparing the ROC curves was not statistically significant (pfive-echo= 0.529, psix-echo=0.558).

Summary: