A growing number of reports refer to MRI of the eye segments and their masses^1,2. Ocular MRI holds the potential to provide guidance during diagnostic assessment and proton beam therapy of uveal melanoma³. Patients scheduled for proton beam therapy undergo implantation of metallic markers which are essential for tumor localization, treatment planning and patient positioning for proton irradiation³. This work examines the MR safety of intraocular tantalum markers used for proton beam therapy of uveal melanoma.

RF power deposition induced heating was studied using electromagnetic field(EMF) and temperature(P_in=1W peak power, t=60s) simulations(Sim4Life V2.0,ZMT,Zurich,Switzerland). E-field coupling was investigated for a tantalum marker implant with a button like geometry(Fig.1a). Alternatively, a disk with the same shape and outer dimensions was employed to generalize and simplify the setup. To approximate the in vivo situation with a given size and shape of the eyeball, the tantalum implants were placed in an agarose phantom at a distance of 15-30mm from its surface. A bowtie electric dipole antenna was used for RF transmission. This approach ensures defined E-field vectors in relation to the implant with the E-field lines being in parallel with the antenna4. Disk rotations between 0° and 90° parallel and perpendicular to the electric field lines were investigated. A high resolution meshbox(resolution (0.05mm)³, size 10x10x10mm³) was defined around the device under investigation(DUI). SAR distribution for various averaging masses(m<1g) was calculated to approach the temperature distribution in the purely SAR dominated linear heating regime. Meeting this criterion, SAR was evaluated locally(SAR_0.075g, cube size (4.2mm)³) and for 1g mass averages(cube size (9.8mm)³).

Magnetic force acting on the marker was investigated with the deflection angle method⁵(Fig.1b). An USP9-0 suture was used to meet the restriction m_suture< 1%m_implant⁵. The holder was positioned at the entrance of the magnet bore(B₀≈1.5T,$$$\nabla$$$B₀≈3.5T/m).

Image artifacts were investigated at 7.0T(Magnetom,Siemens Healthcare,Erlangen,Germany) and 9.4T(Biospec 94/20 USR,Bruker Biospin,Ettlingen,Germany). At 7.0T, a six-channel transceiver array tailored for ophthalmic MRI was employed² with a cylindrical phantom including two spherical holders accommodating an agarose phantom eye(Fig.1c) and ex-vivo pig eyes(Fig.1d). At 9.4T, a 35mm quadrature volume resonator was used. Gradient echo(GRE) and a fast spin echo(FSE) imaging(caption Fig.3) was applied. Three ocular tantalum markers were sutured to the sclera of the pig eye resembling the positioning of the markers in proton beam therapy(Fig.1d).

Peak SAR_0.075g (P_in=1 W peak power accepted at port) obtained for the disk accords very well with max SAR_0.075g observed for the tantalum implant(Fig.1a). SAR_1g clipped peak SAR due to the averaging volume size being significantly larger than the implant (9.8mm vs. 2.5mm;Fig.1). SAR_0.075g showed an exponential decay with increasing distance between the RF antenna and the implant(R²=1;Fig.1b). A linear correlation(R²=1) was found between SAR_0.075g(W) with the implant being present and SAR_0.075g(W/O) without the presence of the implant(Fig.1d). When changing the effective conductive length by rotating the disk perpendicular to the E-field lines a cosine dependence of SAR_0.075g(W)(γ) on the rotation angle was obtained(R²=0.95;Fig.1e,f). Rotation along the E-field lines did not influence the effective conductive length and hence showed no clear angular dependence(Fig.1g). Our simulations demonstrated that SAR_0.075g(W) in presence of the implant can be simply described by SAR_0.075g(W/O) without the presence of the implant: $$SAR_{0.075g(W)}(γ)=1.1·(0.04·cos(2γ)+0.96)·SAR_{0.075g(W/O)}(0°)+0.1$$ Maximum SAR_0.075g(W)=1.9W/kg of the DUI was inferior to maximum SAR_0.075g=2.57W/kg located at the phantom surface. Thermal simulations confirmed a larger temperature increase at the phantom surface(ΔT_max=36.7mK) then at the disk edges(ΔTdisk=18mK) after approximately 2s of heating(Fig.1h). Presence of the disk caused additional heating of δT=1.2mK versus the empty phantom.

Magnetic deflection angle measurement showed no detectable attraction of the implant towards the scanner bore, staying well below the 45° reference point⁵. GRE imaging showed severe susceptibility artifacts around the implant(Fig.3a,b,d,e). FSE images were less affected(Fig.3c,f). Ex-vivo eye imaging(Fig.4) confirmed the strong artifacts in GRE. FSE was less prone to susceptibility artifacts with the artifact towards the vitreous humor barely exceeding the thickness of the sclera.

Minor local increase of RF power deposition was observed for SAR_0.075g(W) but not detectable for SAR_1g. Our studies indicate that intraocular tantalum markers commonly used for proton beam therapy do not constitute a per se contraindication for 7.0T MRI. Strict rejection of subjects with tantalum markers from 7.0T MRI is not justified. Imaging can be conditionally performed after collection of substantial safety information and evaluation of the detailed exposure scenario(RF coil/type and position of implant). This opens the road to high spatial resolution ultrahigh field MRI of patients equipped with tantalum markers who could benefit from MR guided diagnostics, proton therapy beam treatment and response monitoring of ocular masses.