A hybrid, nonlinear programming approach for optimizing passive shimming in MRI

Abstract

BackgroundIn magnetic resonance imaging (MRI), maintaining a highly uniform main magnetic field (B0) is essential for producing detailed images of human anatomy. Passive shimming (PS) is a technique used to enhance B0 uniformity by strategically arranging shimming iron pieces inside the magnet bore. Traditionally, PS optimization has been implemented using linear programming (LP), posing challenges in balancing field quality with the quantity of iron used for shimming.PurposeIn this work, we aimed to improve the efficacy of passive shimming that has the advantages of balancing field quality, iron usage, and harmonics in an optimal manner and leads to a smoother field profile.MethodsThis study introduces a hybrid algorithm that combines particle swarm optimization with sequential quadratic programming (PSO‐SQP) to enhance shimming performance. Additionally, a regularization method is employed to reduce the iron pieces' weight effectively.ResultsThe simulation study demonstrated that the magnetic field was improved from 462 to 3.6 ppm, utilizing merely 1.2 kg of iron in a 40 cm diameter spherical volume (DSV) of a 7T MRI magnet. Compared to traditional optimization techniques, this method notably enhanced magnetic field uniformity by 96.7% and reduced the iron weight requirement by 81.8%.ConclusionThe results indicated that the proposed method is expected to be effective for passive shimming.