This study focuses on the application of the Guidelines¹ issued by ICNIRP (International Commission on Non-Ionizing Radiation Protection) for regulating sensory effects experienced by operators moving through the stray field of the main magnet of an MRI scanner. The research identifies possible violations of the limits and highlights some critical aspects in the implementation of the assessment procedure.The survey², performed via numerical simulations, refers to a 3 T tubular scanner and a 1 T ironless open scanner with vertical axis. The human body is represented through the upper part of the high-resolution anatomical model Duke³, located at different heights above the floor to get three statures, namely 1.62 m, 1.77 m and 1.92 m. These human models are moved near the bore of the two tomographs, along translational/rotational trajectories (11 for the tubular scanner and 6 for the open scanner) including realistic speed profiles. For these 51 exposure situations, both exposure indexes proposed by ICNIRP for “uncontrolled conditions” are computed within the head. During the movements, the maximum variation of the magnetic flux density vector B is evaluated considering the values of the stray field experienced by the head along the trajectory and finally compared to the limit. The exposure index for the motion-induced electric field (EI) is computed through an electromagnetic field formulation, specifically developed and implemented into a numerical code⁴. Since motion-induced electric fields are transient non-sinusoidal signals (whereas the ICNIRP exposure limits are given as a function of frequency), they are processed according to the “weighted peak approach”⁵. This latter technique is implemented and applied both in frequency domain (exploiting the Fourier decomposition) and in time domain (processing the induced signals through a suitable digital filter). The computation of the exposure index is applied globally to all tissues, taken as a whole, and also restricted to the tissues of the central nervous system (CNS) only.

Compliance with the limit for the maximum variation of B is found for all the analyzed cases. Among them, the worst situation is the abrupt head rotation of the shortest human model, performed in front of the bore of the tubular scanner. In this case, the exposure index is lower than 70% of the limit, anyway.

Seven cases exceed the ICNIRP limits for EI (five cases, if the attention is focused on CNS only). They involve the translation toward the bore of the tubular scanner and the abrupt head rotation in close proximity to both scanners. In the worst case, EI exceeds the limit of a factor higher than 2. In general, the highest EI values are found for the shortest human model. Additional tests prove that, for a given path, a speed decrease in general produces a more than linear decrement of EI. The results may undergo significant quantitative variations (increments or decrements, depending on the specific situation) if index EI is computed including/excluding the DC component of the induced signals in its definition (a detail that, in the authors’ opinion, is not well clarified in the Guidelines).

The results suggest that the limit on the maximum variation of B should not be exceeded under normal conditions, even if the speed of motion were increased (because, apart from the shape of the trajectory, it just depends on the level of the stray field). Some violations occur only in additional simulations, where the human models are unrealistically located very close to the internal coils of the scanners. This outcome indicates that compliance with the limit might be broken if the operator placed the head inside the scanning volume.

The analysis of the results for EI indicates a general good agreement between the computations performed in time and frequency domain, even if some non-negligible discrepancies may occur. These differences can be ascribed to the transfer function of the filter, which reproduces the ICNIRP limits with some approximation. However, some intrinsic instabilities in the procedure based on the Fourier decomposition have been also identified⁶. Thus, the two approaches should be considered as complementary. The strong dependence of EI on the DC component of the induced signals calls for specific indications about the treatment of such contribution.

This survey shows that a violation of the limits may occur during MRI practice, but its probability can be reduced by adopting intuitive rules (e.g. reducing the speed of motion and avoiding positions very close to the bore). Some critical aspects in the implementation of the assessment procedure have been highlighted and would deserve clarifications in future revisions of the Guidelines.