Dynamical causal modeling of fMRI time series at 3T: effect of the magnetic field mismatch

Dynamical causal modeling (DCM) is a technique to estimate the connection strengths and modulations between areas in the brain. It is based on an explicit model of brain connections and employs a model for the hemodynamic response to relate neuronal activity and BOLD signal. In SPM, the hemodynamic response is calculated by the balloon model. However, in the implementation of SPM (versions 2, 5 and 8), three parameters (ν₀, the frequency offset at the outer surface of magnetic vessels; r₀, the slope of intravascular relaxation rate as a function of oxygen saturation and ε, the intrinsic ratio of the intravascular to the extravascular signal at rest) are used which explicitly depends on measurements in a 1.5T magnetic field. In many applications of a  DCM analysis with SPM, the data are acquired with a 3T scanner while the model employs parameters based on 1.5T. We call this the magnetic field mismatch.

In this study, we examined the effect of this mismatch using a study on attention acquired on a 3T scanner. We used both the original implementation in SPM and an adapted version in which we used the values for a 3T magnetic field taken from Mildner T et al. A qualitative test of the balloon model for BOLD-based MR signal changes at 3T. Magn Reson Med. 2001: 891-9.  

Application of the original SPM implementation didn't reveale other significant (t-test, uncorrected p-value 0.01) connections or modulations than those obtained with the  parameters for a 3T magnetic field. Furthermore, the significant connection or modulation strengths in our study differed less than 17% except for the input strength which was 3.7 times higher when the magnetic field mismatch was present. 

The magnetic field mismatch in a DCM analysis can give estimates for the connection or modulation strengths which can be greatly over or underestimated but does not necessarily lead to a different interpretation of the underlying model.