Minimizing peripheral nerve stimulation in gradient coil design for head MRI systems using a local field-based linear...

Objective.This work aims to identify a suitable predictor to minimize peripheral nerve stimulation (PNS) in gradient coil design for head magnetic resonance imaging (MRI) systems. The predictor provides an approximate evaluation of PNS thresholds based on local magnetic fields, enabling fast and accurate PNS estimation.Approach.We developed a novel linear threshold predictor to assess PNS thresholds based on the local magnetic fields, and evaluated its performance using realistic human models. The predictor approximated the activating function (AF) by computing a weighted sum of the local field components, and modeled the reciprocal of PNS thresholds as a linear expression of the approximated AF. A series of reference gradient coils were designed to generate a comprehensive dataset, which was used to fit the weighted and linear coefficients in the predictor through a least-squares method. Subsequently, the proposed predictor was incorporated as a linear constraint into a conventional gradient coil design framework to evaluate its effectiveness in predicting PNS thresholds and designing PNS-minimized gradient coils.Main results.The proposed threshold predictor approximated the AF with an estimation error of less than 1%. Maximum prediction errors for PNS thresholds in a head MRI system were 4.7%, 1.7%, and 2.7% for theX-, Y-, andZ-coils, respectively. Incorporating the predictor in the gradient coil design framework could increase PNS thresholds by 119% with a modest 15% increase in coil inductance, while maintaining a maximum linearity deviation of 5%.Significance.The proposed predictor establishes an approximate relationship between local magnetic fields and PNS thresholds, providing an accurate and efficient tool for evaluating PNS thresholds and enabling the design of PNS-minimized gradient coils for safer head MRI systems.