Pulmonary MRI in Interstitial Lung Disease: A Systematic Review and Meta-Analysis

Magnetic resonance imaging (MRI) has emerged as a radiation-free alternative for evaluating interstitial lung disease (ILD), but its diagnostic performance relative to computed tomography (CT) remains variably reported.

To evaluate the diagnostic performance of pulmonary MRI for ILD detection and identification of key imaging hallmarks, using CT as the primary reference standard.

Systematic review and meta-analysis of observational studies.

Twenty-five studies including 959 participants were included in the qualitative synthesis; nine studies (n = 473) were eligible for quantitative meta-analysis.

MRI field strengths ranged from 0.55 to 3.0 T and included ultrashort echo-time (UTE), zero echo-time (ZTE), T2-weighted fast spin-echo, and gradient-echo-based sequences with slice thickness ranging from 1 to 20 mm.

MRI performance was evaluated against CT for global ILD detection and imaging hallmarks including ground-glass opacities, nodules, traction bronchiectasis, reticulations, and honeycombing.

Pooled sensitivity and specificity were estimated jointly using a bivariate hierarchical model of Reitsma et al.; heterogeneity was assessed with I² statistics (p < 0.10). Leave-one-out sensitivity analyses were performed.

Pooled sensitivity was 0.91 (95% CI: 0.84-0.95; I²: 42.1%) and pooled specificity was 0.86 (95% CI: 0.73-0.93; I²: 62.6%) and summary SROC AUC of 0.939. In a predefined subanalysis restricted to UTE/ZTE thin-slice protocols (four studies), sensitivity was 0.93 (95% CI: 0.87-0.97; I²: 0.0%) and specificity was 0.92 (95% CI: 0.78-0.98; I²: 51.3%).

Pulmonary MRI demonstrates high diagnostic performance for ILD detection, with results suggesting protocol-dependent performance and improved consistency under optimized UTE/ZTE thin-slice acquisitions.

3.

Stage 2.