Obstructive sleep apnea (OSA) is characterized by repetitive upper airway (UA) collapse during sleep. Invasive endoscopy has shown that airway cross sectional area change during Mueller maneuver (MM) is correlated to the severity of OSA¹. MM is a voluntary effort with poor reproducibility. Continuous positive airway pressure (CPAP) can produce similar and well-controlled positive pressure, and is the most widely used therapy for OSA². The direct effects of CPAP on soft tissues surrounding the upper airway have been extensively studied using static MRI³. However, the underlying mechanisms of airway response to pressure change remains unknown. In this study, we perform real-time multi-slice MRI during rapid changes to the CPAP pressure level, and for the first time, fully resolve the soft tissue response to pressure change.

Acquisition: Experiments were performed on a clinical 3T MRI (GE EXCITE HDxt) with a 6-channel carotid coil. Physiological signals including facemask pressure, respiratory effort bellow displacement, oxygen saturation and heart rate were simultaneously collected. The facemask was connected to a CPAP machine (Philips Respironics System One) in the MRI control room. The maximum pressure level in the facemask was set to a pre-titrated value that maintained patent airway, determined during an overnight sleep study using standard polysomnography. During each study, the CPAP level in the facemask was alternated between the maximum value and 4cm H₂O. Note some positive pressure is required, due to resistance in the long tube between CPAP and facemask. The imaging sequence, described in Ref⁴ utilized a simultaneous multi-slice golden angle radial CAIPIRINHA fast gradient echo acquisition, providing 1mm in-plane spatial resolution, 4 simultaneous slices (2-retroglossal 2-retropalatal), with 96ms temporal resolution.

Image analysis: A semi-automated region growing algorithm⁴ was used to segment the airway in each 2D slice. The lateral and anterior-posterior dimensions of the patent airway were calculated using the segmented images, by subtracting the left-right and anterior-posterior extents of the airway, respectively.

Fig.1 contains representative images from a patient with OSA (15 yrs., male, AHI 11.6), with the top row at 4cm H₂O, and bottom row at 11cm H₂O, covering from soft palate to epiglottis (left to right). Fig.2 contains the temporal changes in airway area for all four slices during sudden pressure rise at t=1 sec (4 to 11 cm H₂O) and drop at t=16sec (11 to 4 cm H₂O). The pressure remained at 11cm H₂O in between these two time points, however, the airway area gradually changed during that timespan, and with tidal breathing, indicating a highly dynamic process. Fig.3 shows the temporal change in airway dimension along the lateral and anterior-posterior directions, corresponding to the right two columns in Fig.1. Airway change in our (n=4) cohort was mediated primarily by lateral expansion. The collapse pattern has been observed for all 4 subjects to date, conforms to similar observation in Ref³.Pathological mechanisms underlying OSA have been studied by measuring physiological changes during immediate drops in CPAP⁵. The proposed technique allows one to fully resolve the dynamics secondary to sudden pressure change, therefore has the potential to provide visual and quantitative information for pathological assessment. Furthermore, simultaneous multi-slice acquisition allows identification of the more responsive site to CPAP change. For example, the tissues in the right two columns (retroglossal) deform more than the other two columns (retropalatal), and can be positively identified as the more compliant sites for this particular subject. A previous study using static MRI³ showed airway area enlarged with progressively increased pressure. However, the fully resolved dynamics (highlighted area in Fig.2) reveals that airway area depends on many factors in addition to the pressure level. We demonstrate a novel experiment that captured the upper airways’ instantaneous response to pressure changes. By performing real-time multi-slice MRI during the application of CPAP, we demonstrate that airway area change is dynamic and depends on factors including pressure level, airway section and previous muscle tone status.