Quantitative magnetic resonance imaging (qMRI) in axial spondyloarthritis

Abstract

Imaging, and particularly MRI, plays a crucial role in the assessment of inflammation in rheumatic disease, and forms a core component of the diagnostic pathway in axial spondyloarthritis. However, conventional imaging techniques are limited by image contrast being non-specific to inflammation and a reliance on subjective, qualitative reader interpretation. Quantitative MRI methods offer scope to address these limitations and improve our ability to accurately and precisely detect and characterise inflammation, potentially facilitating a more personalised approach to management. Here, we review quantitative MRI methods and emerging quantitative imaging biomarkers for imaging inflammation in axial spondyloarthritis. We discuss the potential benefits as well as the practical considerations that must be addressed in the movement toward clinical translation of quantitative imaging biomarkers.

Figure 1.

A flowchart of the steps involved in the qMRI pipeline (with chemical shift encoded MRI (CSE-MRI) as an example) and comparison with conventional MRI acquisition.

Figure 2.

Active inflammation at the left sacroiliac joint detected on the short inversion time inversion recovery (STIR) image (top) is detectable on the qMRI diffusion-weighted image (bottom). The arrows indicate bone marrow oedema (raised apparent diffusion coefficient, ADC value).

Figure 3.

Fat fraction (FF) mapping detects active inflammation and bone marrow oedema (figure reproduced from Bray et al). Active inflammation detected on the STIR image (a) is detectable on the fat fraction map as a reduction in fat fraction (b). The fat fraction map can also detect fat metaplasia (d), a form of structural damage which is conventionally identified using T1 weighted images (c).

Figure 4.

DCE-MRI of two patients with axSpA demonstrating active (a,b) and inactive (c,d) disease (figure reproduced from Zhang et al). (a) active inflammation is detected as an area of enhancement (arrows) in the right SIJ on the contrast-enhanced T1-weighted MR image. (b) K^trans (forward volume transfer constant) colour map. K^trans is a measure of capillary permeability which is calculated by measuring the accumulation of contrast in the extravascular-extracellular space. The map shows high K^trans values (K^trans  = 1.975) in the inflamed right SIJ (arrows). (c) No significant enhancement on the enhanced T1 image. (d) K^trans values were relatively low in bilateral sacroiliac joint; multiple ROIs were placed on the articular surface and the average K^trans was 0.388.

Figure 5.

IVIM DWI in a patient with axSpA (figure reproduced from Zhao et al). On the Ds (pure molecular diffusion) map, ƒ (perfusion fraction) map, and Df (perfusion-related diffusion) map, ROIs are placed in the juxta-articular bone marrow. IVIM diffusion decay curves can be plotted with b value (x axis) against log plot of signal intensity (y axis) based on a monoexponential model (blue line) and biexponential model (red line). For this ROI, IVIM DWI signal intensity decay shows a nonlinear relationship (top right).

Figure 6.

Improving the objectivity of ROI placement using the BEACH tool (figure reproduced from Bray et al (with fat fraction (FF) maps as an example). With this tool, the user identifies the joint line (a) and places a linear region of interest to define the joint (b), as well as anchor lines to define the angle made between the joint and the edge of the bone (see reference for details). The tool then automatically propagates a polygonal ROI onto the subchondral bone of interest (c). This region also includes normal bone, but the BEACH tool uses histographic analysis to target areas of maximal abnormality.

Figure 7.

Semiautomated analysis using deep learning (figure reproduced from Hepburn et al). In the method proposed by Hepburn et al, areas of inflammation (shown as hyperintense on STIR images, left column) are segmented by an algorithm combining deep learning with intensity-based thresholding (segmented disease in middle (shown by arrows) and right column). In this example, the volume of abnormal tissue is shown to markedly reduce after treatment.