INTRODUCTION

Signal voids caused by metallic needles pose significant visualization, guidance and monitoring challenges in many interventional MRI (iMRI) applications including biopsies^1-3, ablations^4,5 and brachytherapy^6,7. In earlier work, we presented the design and simulation of active shims for metallic probes and demonstrated correction of the ΔB₀ and reduction of the signal void around a titanium needle⁸. In this work, we present the actual fabrication and correction of the signal void artifact around a titanium needle at 3T. Artifact correction is shown to extend across imaging with multiple contrasts and orientations.

METHODS

Simulations
The needle induced field, shim coil paths, shim fields and field shimming were simulated in Matlab. ΔB₀ around a 4/3 mm OD/ID, hollow Titanium needle with a 30^o single-sided bevel (Volume Susceptibility: 𝟀 = 182*10-6 )⁹ in Water (𝟀 = -9.05*10-6 )⁹ was modeled at 3T using Fourier-analysis field modeling¹⁰. A set of 18 orthogonal 2-coil shim paths were modeled for fields produced and shimming^8,11, from which a pair consisting of a ½ turn loop angled at the bevel angle (CN90) and a split wire loop (CN0) was selected for best predicted performance (Figure 1).

Needle and Shim Insert Fabrication
A 4/3 mm OD/ID, 15 cm hollow titanium tube was used to fabricate the needle. For mounting the shim coils, a 2.5 mm, 15 cm former with a bevel and orthogonal wire slots was designed in Solidworks (Dassault Systemes, USA) (Figure 2a). The insert was 3D printed on a ProJet 3500 printer (3D Systems Inc, USA). 26-gauge enameled copper wire was placed into the slots by hand and insulated using a single layer of kapton tape. The insert was then slid inside the needle with the bevels aligned with each other (Figure 2b).
The needle was tightened into a central bore hole of a 3D printed holder with a compartment for shim wire connections (Figure 2c). Shim wires were connected to 20 feet long twisted cables for current supply (E3631A, Keysight Technologies, USA, 6V/±5A, 25V//±1A). A pair of high impedance 127 MHz RF chokes¹² were placed in line along each current path to minimize RF pulse induced currents.

Shimming Experiments
All experiments were performed on a Philips 3 Tesla scanner with a 2-channel body RF transmit and 16 channel neuro-vascular receive coil. Two 15 cm diameter cylindrical phantoms with domes that allowed needle insertions at different angles were designed and 3D printed (Figure 2d). The objectives of the needle shimming experiment were to a) demonstrate and quantify the recovery of lost signal around the needle at different orientations and b) show artifact reduction in different imaging sequences commonly used in iMRI applications.
The currents necessary to shim the needle at four different orientations were simulated: i) 0,0,0 ii) 90,0,0 iii) 0,18,30 and iv) 20, -30,0 degrees. The needle containing the shim coils was then inserted into the water phantom through the guide holes in these orientations and imaging was performed with 3D GRE with fieldmapping, 3D FSE, 2D single slice balanced SSFP and single slice multishot EPI sequences for each (Parameters in Figure 3). Images were acquired in the planes parallel and perpendicular (for 3D GRE) to the needle bevel. For each orientation, images were acquired without and with the pre-calculated shim currents supplied to the two coils simultaneously.

Shim Performance Evaluation
Shim performance was assessed by estimating the volume of the signal void around the needle for the GRE images. Voxels with less than 25% of the mean signal from a reference slice away from the needle were identified as the signal void and the total volume of these voxels in mm³ was calculated. The percentage of lost signal recovered by shimming was also calculated.

RESULTS

Simulations predicted the need for a) 1.17A, 0 A b) 0A, -1.15A c) 0.97A, 0 A and d) 0.819A, -0.29A in the CN0 and CN90 coils for the four needle orientations. Figure 4 shows parallel, orthogonal and cross-section slices from the 3D GRE acquisitions. Significant recovery of the lost signal around the needle is evident in all orientations with shimming. The volume of the signal void reduced from 8097, 8047, 3324 and 5072 mm³ without shimming to 2510, 2816, 1428, 1995 mm³ with shimming respectively for the four orientations. This represented recovery of 69.0%, 64.9%, 57.1% and 60.6% of the lost signal. This value would in-fact be higher if the volume of the needle could be exactly estimated and accounted for. The fieldmaps show the correction of the underlying ΔB₀. Without shimming, there is significant field wrapping around the needle due to extremely high ΔB₀, which is eliminated with shimming. Figure 5 shows Balanced SSFP, TSE and Multishot EPI images without and with shimming. Recovery of signal is seen to be consistent across different contrasts due to the correction of the underlying ΔB₀.

DISCUSSION

Significant recovery of lost signal around a metallic needle extending across orientations and imaging contrasts is shown with active shimming with pre-calculated shim values. Minimal artifact needles can find many uses in iMRI including for improving target visibility in biopsies, monitoring of therapy in ablations, delineation of targets and at risk structures in brachytherapy among others.

Acknowledgements

This work was supported by NIBIB R21EB025258. I would like to thank Dr William A Grissom, Gary Drake and Dr Megan Poorman help and for advice on field modeling and 3D printing.