Classification Scheme of Heating Risk during MRI Scans on Patients with Orthopaedic Prostheses

Abstract

Due to the large variety of possible clinical scenarios, a reliable heating-risk assessment is not straightforward when patients with arthroplasty undergo MRI scans. This paper proposes a simple procedure to estimate the thermal effects induced in patients with hip, knee, or shoulder arthroplasty during MRI exams. The most representative clinical scenarios were identified by a preliminary frequency analysis, based on clinical service databases, collecting MRI exams of 11,658 implant carrier patients. The thermal effects produced by radiofrequency and switching gradient fields were investigated through 588 numerical simulations performed on an ASTM-like phantom, considering four prostheses, two static field values, seven MR sequences, and seven regions of imaging. The risk assessment was inspired by standards for radiofrequency fields and by scientific studies for gradient fields. Three risk tiers were defined for the radiofrequency, in terms of whole-body and local SAR averages, and for GC fields, in terms of temperature elevation. Only 50 out of 588 scenarios require some caution to be managed. Results showed that the whole-body SAR is not a self-reliant safety parameter for patients with metallic implants. The proposed numerical procedure can be easily extended to any other scenario, including the use of detailed anatomical models.

Keywords: MRI heating risk; MRI safety; gradient-induced heating; orthopaedic implants; radiofrequency-induced heating.

Figure 1

Prosthesis models used in the in silico heating evaluation.

Figure 2

Positions of the human body with respect to the RF body coil and the GC (dark grey rectangle) for the considered imaging zones. The simulated implant positions are represented by the coloured rectangles and the phantom is represented by the dashed white rectangle.

Figure 3

Chromatic map of maximum admissible configuration RF SAR index ${\xi }_{TR,Max}$ as a function of the RF stress index and of the sequence duration for hip and shoulder implants. The corresponding points in the ψ_A-T_seq plane of the used sequences are reported. The dashed line represents a sequence duration equal to T_IEC/2. Below this value, ${\mathsf{\xi }}_{10\mathrm{g},S}^{*}$ becomes independent of the sequence duration.

Figure 4

Configuration index ${\xi }_{10g,C}$ for the most stressed situations at 1.5 T (a) and 3 T (b). The maximum values of this index (${\xi }_{10g,Max}$) admissible for condition (1) and (2) associated with the TrueFISP, T1 FSE, and T2 FRFSE sequences are reported as solid lines (condition 1) and dotted lines (condition 2). For compatibility reasons, the values of ${\xi }_{10g,C}$ for knee are fictitiously halved.

Figure 5

Radiofrequency (RF) and gradient coil (GC)-heating stress evaluation for the scenarios of risk tier 3 or tier 2. Tiers 3, 2 and 1 are associated with red, yellow and green colours, respectively.

Figure 6

Relative variation in the estimated SAR quantities against relative variation in the RF pulse duration when the time–bandwidth product is kept constant. The result of this sensitivity analysis is independent of the reference sequence.