Although synovitis detection is optimal with contrast-enhanced, high field magnetic resonance imaging (MRI), the increased accessibility of power Doppler ultrasound (PDUS) has meant that it is more frequently utilised in the clinical setting. However, with the advent of low field extremity MRI (eMRI) and its practical advantages,1 there is a need to understand what is the most sensitive tool for routine clinical use. eMRI has been evaluated with respect to its ability to assess rheumatoid arthritis erosions in comparison with both high field MRI and radiography.2^–4 There is relatively little published work, however, on eMRI’s ability to assess synovitis in comparison with high field MRI,4^–7 and three of these reports used intravenous contrast, which further limits feasibility. Only one group have published data comparing eMRI with PDUS, which may be a more clinically relevant comparison, and again this study used contrast enhanced eMRI.8

In order to assess synovitis without contrast injection, low field eMRI employs a single short τ inversion recovery (STIR) pulse sequence to suppress the MRI signal from fat. Such sequences on low field machines are, however, limited by a low signal to noise ratio (SNR) that reduces the quality of the image, and the STIR sequences also vary in quality between different types of eMRI machine.9

The aim of the current study was to compare synovitis detection with eMRI without contrast injection and PDUS.

A total of 254 eMRI and PDUS images of the most symptomatic metacarpophalangeal 2–3, radio-carpal, inter-carpal and distal radio-ulnar joints were obtained, using both early (n = 32 subjects, median disease duration 3 months) and long-standing rheumatoid arthritis cohorts (n = 32 subjects, median disease duration 7 years). Using a MagneVu MV1000 (MagneVu, Carlsbad, CA, USA) machine, the standardised protocol for the STIR pulse sequence was as follows: TR/TE/TI, 100/24/50 ms; field of view 50×75 mm ×15 mm, four excitations, 0.625 mm individual slice thickness (Z plane), 1 mm coronal (X and Y in-plane) resolution. Images were subsequently scored for the presence of synovitis using the RAMRIS (Rheumatoid Arthritis Magnetic Resonance Imaging Score) score, which grades synovitis on a semi-quantitative 0–3 scale.10 On the same day, PDUS was performed using a Philips HDI 5000 machine with a 15–7 MHz hockey stick probe, and joints were scored for PD synovitis using a semi-quantitative scoring (0–3) method. The US images were obtained by two experienced ultrasonographers with an inter-reader weighted κ of 0.72. Percentage exact agreement (PEA) was calculated on a joint-by-joint basis. Dichotomised scores (where grades 0–1 and 2–3 were grouped together for both imaging modalities) were used in order to produce sensitivity and specificity data for eMRI, where PDUS was used as the gold standard. The results are presented in table 1.

eMRI using the MV1000 demonstrated poor sensitivity in comparison with PDUS, although specificity was good. Studies comparing eMRI and high field MRI synovitis assessment have demonstrated better sensitivity but used a different eMRI machine (the Esaote C scan)4^–7 and several studies used intravenous contrast4 5 7 and are therefore not directly comparable. It is not possible to use intravenous contrast with the MV1000 as it is not able to resolve the images quickly enough, ie, while the contrast is still in the region of interest. It should also be noted that PDUS is an examiner-dependent technique that may affect the choice of imaging modality employed in clinical practice. In summary, using the MV1000 without contrast is less sensitive at detecting synovitis than PDUS, but care should be taken to recognise the differing capabilities of various eMRI machines.