Body gradient performance is strongly limited by peripheral nerve stimulation (PNS) thresholds, which is an impediment to higher performance MRI in many fields. To overcome this limitation, we have designed and built a high performance gradient with a linear region to just accommodate the human brain and upper C-spine (22cm transverse, 19cm longitudinal), capable of 120 mT/m and 1200 T/m/s using presently available 1MVA gradient drivers. This head gradient, denoted “H3”, uses a short, symmetric, folded design, compact enough to be rapidly inserted into a standard clinical scanner without removing the body gradient or RF coil[1-4]. We anticipated the restricted linear region would lead to PNS thresholds substantially higher than for body gradients[5], and possibly also previous head gradients. We present the first measured PNS thresholds for our novel high performance head gradient.

PNS thresholds were measured in volunteers (N=12) outside the main magnetic field under an IRB-approved protocol. The absence of B₀ should not affect PNS, and led to quiet, minimal vibration measurements, aiding detection of subtle stimulation onset. The gradient was connected to standard amplifiers (660A, 1650V peak) and controlled using the scanner console (Discovery MR750, GE Healthcare, Waukesha WI).

Waveforms (Fig. 1) consisted of trains of trapezoidal lobes 64 pulses long with a fixed flat-top time of 1ms[5]. The peak to peak rise time (τ) was varied between 100ms and 1600ms and gradient excursion (ΔG) was initialized at a value known not to stimulate and was then slowly increased with a 2s pause between increments. When the subject reported the first definite PNS, the waveform pulsing was stopped, and the threshold (τ and ΔG) recorded. A plot of ΔG vs. τ reveals the linear PNS threshold curve described by ΔG_min and SR_min[6].

The resulting threshold data were analyzed using logistic regression[5] to extract population mean values for ΔG_min and SR_min. The physiological PNS parameters, chronaxie (τ_c) and rheobase (dB/dt_r), were derived from these[1]. Only those rise times for which at least 25% of the population stimulated were included in the analysis[1].

Stimulation thresholds for the individually-driven X and Y axes are shown in Figure 2, with the logistic mean and standard deviation indicated for each τ, as well as linear fits to these population mean values. Figure 3 indicates the oblique XY (simultaneous X and Y) axis, has higher PNS and hardware thresholds than either of the individual axes. Reference PNS thresholds and hardware limits for existing head[5,7,8] and body[5] gradient coils are shown, whenever corresponding measurements were available.

The PNS threshold parameters for all axes are summarized in Table 1. For our H3 coil, X and Y axes produced stimulations for most volunteers enabling precise estimates of ΔG_min and SR_min. PNS was felt primarily around the face for the x-axis, and around the neck, shoulders and ears for the y-axis. Two subjects reported sensations in their fingers which did not depend on hand position. The H3 Z-axis did not stimulate enough volunteers to establish PNS thresholds. For the oblique axes. ∆G values are defined as the vector magnitude[5].

Our gradient hardware capabilities are at least 50% higher (strength and slew rate) than existing head gradients. Importantly, the PNS thresholds for our head gradient are greater than existing head gradient coils, and 3-4 times higher than existing body gradients! We attribute these major increases in PNS thresholds to the smaller linear region length (19cm) for our head gradient compared to existing head gradients (26cm for HG2[8]).

For previous head gradient coils (HN1, HG2), the x-axis had the lowest PNS threshold[5,7]. This was not the case for our H3 head gradient, for which the Y gradient had the lowest threshold (comparable to the x-axis threshold of HG2). The H3 x-axis threshold was 60% higher (ΔG_min) than the H3 Y-axis. We have not established the underlying mechanism for this “swapped” X-Y PNS threshold behaviour of H3 compared to other head gradients, but speculate that it is a result of the folded design of H3.

H3 is capable of significantly higher strengths and slew rates than have been achievable with previous head only coils; this has allowed us to measure stimulation thresholds on two out of three single axes, whereas most previous PNS studies have only been able to produce stimulation on one single axis and have had to rely on oblique axis drive to reliably measure stimulation thresholds[5,7].

We have characterized the PNS performance of a novel insertable head gradient coil, and found these PNS thresholds to be substantially higher than for previously characterized head and body gradient coils.