Off-resonance based assessment of metallic wear debris near total hip arthroplasty

Abstract

Purpose: The presence of metallic debris near total hip arthroplasty can have a significant impact on longitudinal patient management. Methods for magnetic resonance imaging-based quantification of metallic debris near painful total hip replacements are described and applied to cohorts of symptomatic and control subject cases.

Methods: A combination of metal artifact reduction, off-resonance mapping, off-resonance background removal, and spatial clustering methods are utilized to quantify off-resonance signatures in cases of suspected metallosis. These methods are applied to a cohort of symptomatic hip arthroplasties composed of cobalt-chromium alloys. Magnetostatic simulations and theoretical principles are used to illuminate the potential sources of the measured off-resonance effects. Reported metrics from histological tissue assays extracted during surgical revision procedures are also correlated with the proposed magnetic resonance imaging-based quantification results.

Results: The presented methods identified quantifiable metallosis signatures in more than 70% of the symptomatic and none of the control cases. Preliminary correlations of the MR data with direct histological evaluation of retrieved tissue samples indicate that the observed off-resonance effect may be related to tissue necrosis.

Conclusions: Magnetostatic simulations, theoretical principles, and preliminary histological trends suggest that disassociated cobalt is the source of the observed off-resonance signature. Magn Reson Med 79:1628-1637, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Keywords: metal implants; metallosis; susceptibility quantification; total hip arthroplasty.

Figure 1

Results from one iteration of Monte Carlo particle off-resonance simulation for CoCr particulate matter with 150 nm resolution spread across a cubic volume of 10 μm edge-length at 1.5T. (A) displays the distribution of particles, while (B) presents the off-resonance values induced in non-particulate (i.e., tissue) grid positions. The results of the Monte Carlo simulation, run with 1000 iterations, is shown in (C). It is clear that the mean of the expected Larmor frequency offset is negligible

Figure 2

Demonstration of MRI mScore computation process. Large field of view MAVRIC SL image (A), indicated analysis region (B), identified off-resonance cluster superimposed on magnitude image (C). Full field of view (D) and zoomed (E) 3D-MSI off-resonance maps. Background extracted off-resonance field in the analysis region (F).

Figure 3

Case studies of six symptomatic subjects with suspected metallosis. Localized background-removed off-resonance maps (bottom row) and correlated magnitude MAVRIC SL 3D-MSI images (top row) are displayed. Computed mScores are also displayed for each case.

Figure 4

Importance of off-resonance mapping method. The conventional center of mass approach is prone to ghosting artifacts in far off-resonant bins, which displaces the off-resonance center frequency estimate for a given voxel (dashed line). The spectral model suppresses these bins from the center frequency estimate (solid line) and provides a map without false positive off-resonance pockets. Application of the center of mass approach yielded a 25% false positive rate in the control analysis, whereas the spectral modeling approach resulted in no false positives.

Figure 5

(A) Distribution of mScores computed for cohort of subjects with symptomatic total hip replacements. (B) Plot of histological necrosis score (Natu) versus mScore. Linear regression analysis of the relationship yields p = 0.08. Though the trend does not reach statistical significance (p < 0.05), as indicated in Table 0.1, it shows shows the best correlation of all measured histology metrics.