Comparison of bi-exponential and mono-exponential models of diffusion-weighted imaging for detecting active sacroiliitis in ankylosing spondylitis

Abstract

Background

There has been a growing need for a sensitive and effective imaging method for the differentiation of the activity of ankylosing spondylitis (AS).

Purpose

To compare the performances of intravoxel incoherent motion (IVIM)-derived parameters and the apparent diffusion coefficient (ADC) for distinguishing AS-activity.

Material and Methods

One hundred patients with AS were divided into active (n = 51) and non-active groups (n = 49) and 21 healthy volunteers were included as control. The ADC, diffusion coefficient (D), pseudodiffusion coefficient (D*), and perfusion fraction (f) were calculated for all groups. Kruskal–Wallis tests and receiver operator characteristic (ROC) curve analysis were performed for all parameters.

Results

There was good reproducibility of ADC/D and relatively poor reproducibility of D*/f. ADC, D, and f were significantly higher in the active group than in the non-active and control groups (all P < 0.0001, respectively). D* was slightly but significant lower in the active group than in the non-active and control group (P = 0.0064, 0.0215). There was no significant difference in any parameter between the non-active group and the control group (all P > 0.050). In the ROC analysis, ADC had the largest AUC for distinguishing between the active group and the non-active group (0.988) and between the active and control groups (0.990). Multivariate logistic regression analysis models showed no diagnostic improvement.

Conclusion

ADC provided better diagnostic performance than IVIM-derived parameters in differentiating AS activity. Therefore, a straightforward and effective mono-exponential model of diffusion-weighted imaging may be sufficient for differentiating AS activity in the clinic.