Comparison of ultrasound and dynamic MRI for the measurement of diaphragmatic excursion: A prospective single-center study

Abstract

Objectives Ultrasound (US) measurements of diaphragmatic excursion (DE) are widely used to provide a non-invasive assessment of the diaphragmatic function at the bedside, especially in intensive care. However, this measurement has never been validated against a less operator-dependent technique such as MRI. Dynamic MRI is the only imaging modality that creates a four-dimensional reconstruction of the diaphragm. The primary objective of this study was to assess the agreement between DE obtained using dynamic MRI with those obtained using ultrasound. The secondary objectives were to define DE thresholds for the diagnosis of DD using MRI and to compare the performance of US and MRI to diagnose DD. Methods Prospective single-center study in which consecutive outpatients referred for a dynamic thoracic MRI were included. This study was conducted at a university hospital in Paris, where there was daily access to ultrasound (US) and extensive expertise in diaphragmatic MRI The DE of each hemi-diaphragm was measured sequentially using ultrasound and MRI in random order, during spontaneous breathing (SB) and forced inspiration (FI) by independent observers blinded to each other. We analyzed the agreement between DE obtained using US and MRI for each hemi-diaphragm. Results We enrolled forty-five patients, aged 58 ± 36 years, of which twenty-eight (68%) had a confirmed DD. During SB, the mean bias for DE measurement was −3.8 mm, 95% CI [−7.1; −0.6] for the left hemi-diaphragm, and 1.0 mm, 95% CI [−3.5; 5.5] for the right hemi-diaphragm. Limits of agreement (millimeters) were [−25; 17] on the left side, and [−28; 30] on the right side. MRI threshold values for DE defining dysfunction were 11 mm for quiet SB, and 38 mm for FI. These thresholds had a sensitivity of 77.7% and a specificity of 77.4% during SB, with an AUC of 0.86. Conclusion US and MRI provide different values for DE, probably because the measurements were not obtained exactly at the same localization. Nevertheless, diagnostic performances of MRI and US to recognize DD appeared comparable.