AGREEMENT AND PRECISION OF FEMOROTIBIAL CARTILAGE TRANSVERSE RELAXATION TIME (T2) OBTAINED AT FOUR DIFFERENT MAGNETIC FIELD STRENGTHS

Abstract

Spin-spin (transverse) relaxation time (T2) is considered to reflect cartilage composition, such as hydration, collagen content and orientation. It can be computed from multi-echo spin-echo (MESE) MRI and is associated with age, sex, and structural joint pathology. Cartilage T2 is dependent on the magnetic field strength. Lower field strength is characterized by lower signal to noise ratios (SNR) compared to standard 1.5Tesla (T) and 3T systems, resulting in longer acquisition times, while access to ultra-high field 7T MRI is still limited. A direct comparative evaluation of cartilage T2 across different field strengths has not been performed to date. To analyze the agreement and the intra-reader test-retest precision of laminar femorotibial cartilage T2, acquired from four different scanner models with 0.55T,1.5T, 3T, and 7T. Six healthy volunteers (5 female, age 21.5±3.5 y) without knee pain had two MESE MRIs acquired each, with repositioning of the knee between test and retest scans, using a 1) MAGNETOM Free.Max 0.55T (acquisition matrix (AM) 304 × 248, slice thickness [ST] 3mm, TR 3020ms, 6 echoes from 16.1 to 96.6ms, acquisition time (AT) 14 min 56 sec) 2) MAGNETOM Sola 1.5T (AM 384 × 384, ST 3mm, TR 2700ms, 7 echoes from 11.7 to 81.9ms, AT 10 min 31 sec) 3) MAGNETOM Vida 3T (AM 384 × 384, ST 3mm, TR 3210ms, 9 echoes from 11.5 to 103.5ms, AT 8 min 17 sec) 4) MAGNETOM Terra 7T (AM 384 × 384, ST 3mm, TR 4490ms, 9 echoes from 8.1 to 72.9ms, AT 8 min 22 sec) (Fig. 1). The field of view was 12 × 12cm for all MESE MRIs. Segmentation of the medial and lateral femorotibial compartment cartilage (MFTC/LFTC) was performed manually by an expert reader; the T2 values were computed for each voxel using custom software. The cartilage layers were divided into the top 50% (superficial [sf]) and bottom 50% (deep) layer, based on the local distance between the cartilage surface and bone interface. The test-retest precision across the entire femorotibial joint (FTJ), the MFTC, and the LFTC was measured using the root mean square coefficient of variation (RMS CV in %). The absolute values of laminar T2 were compared between field strengths using ANOVA, with Bonferroni-Dunn post-hoc correction. The RMS CV across the FTJ was lowest at 3T (0.9%/1.4% sf/deep) and greatest at 0.55T (3.2%/4.0%). The RMS CV was 1.3%/3.4% at 1.5T and 2.3%/1.4% at 7T. Sf RMS CV at 0.55/1.5/3/7T was 3.9/1.7/1.0/3.5% for the MFTC, and 3.0/1.7/1.2/2.7% for the LFTC. The deep layer RMS CV was 5.4/3.4/3.1/2.6% for the MFTC and 5.0/4.1/1.1/2.9% for the LFTC. Sf and deep layer T2 differed significantly between field strengths for the FTJ, the MFTC, and the LFTC (all p<0.001). Both sf and deep layer T2 across the FTJ, the MFTC and also the LFTC were shortest at 7T and longest at 0.55T (Fig. 2). Interestingly, T2 was shorter at 1.5T than at 3T, but this difference reached statistical significance only for the deep layer in the FTJ and in the MFTC (Fig. 2). Based on the protocols implemented in this study, laminar cartilage T2 systematically varied with magnetic field strength. The shortest values were observed at 7T, and the longest at 0.55T. These findings suggest that cartilage T2 measurements in multicenter studies are not directly comparable when different field strengths (potentially also different manufacturer and scanner models) are included. The test-retest precision error was lowest at 3T and highest at 0.55T, suggesting that, in clinical studies, measurements should be performed at 3T. However, test-retest precision is highly dependent on the protocol and might be improved for lower field strengths, applying different acquisition parameters (e.g. longer acquisition times and higher resolution). Implementation on the 7T magnet may require further improvement to leverage the full potential benefit of the higher field strength.