To develop and demonstrate a real-time imaging method with adequate spatiotemporal resolution and coverage for assessing upper airway collapsibility in sleep apnea.Obstructive sleep apnea (OSA) is characterized by repetitive upper airway (UA) collapse during sleep. UA compliance, defined as the ratio of UA cross-sectional area and pressure, has been used to measure airway collapsibility¹. Single-slice compliance measurement has been performed using real-time imaging², however extended spatial coverage is essential in order to characterize collapse pattern. Here we present a method for simultaneous multi-slice compliance measurement based on sparse golden-angle radial CAIPIRINHA³.

Experiment Setup: experiments were performed on a clinical 3T scanner (EXCITE HDxt, GE) using a 6-channel carotid receive coil. Physiological signals including facemask pressure, abdomen bellow displacement, oxygen saturation and heart rate were simultaneously recorded to determine wakefulness/sleep. The mask was occasionally occluded to generate enough negative pressure for measuring compliance. Five adolescent OSA patients were studied and three of them fell asleep. Three adult OSA and four adult healthy volunteers were also studied during wakefulness.

Data acquisition: to image N slices, a total of N unique multi-band RF pulses were applied alternatively. The n^th pulses was designed such that the phase difference between adjacent slices was $$$2\pi n/N, n\in [0,N-1]$$$. Continuous radial acquisition with 1/N golden-angle increment was used. Imaging parameters were: radial FLASH, 5˚ flip angle, 7mm/3mm slice thickness/spacing, 200 samples per readout, FOV 200x200mm², TR 4ms.

Reconstruction: 24-32 spokes were used to reconstruct each temporal frame without view-sharing, which led to 96-128ms temporal resolution. Each slice was reconstructed separately by iteratively minimizing the cost function $$f_i=\sum_j ||ES_{ij}m_i-P_ik_j||_2^2+\lambda_i||\phi m_i||_1, i\in[1,N], j\in[1,N_c] $$ where i,j are the slice and coil index, P_i is the conjugate of RF phase cycling pattern, E is the inverse gridding operator, S_ij is the coil sensitivity map, k_j is the acquired k-space data, $$$\phi$$$ is the finite temporal difference, and m_i is the image to be solved.

Postprocessing: the airway was segmented in each frame using a semi-automated region-growing algorithm². The airway area was normalized by the maximum cross-sectional area among all slices during tidal breathing, in order to enable inter-subject comparison. For each slice, all data from one occluded breath were used to perform a linear regression (airway area vs pressure), from which the compliance (line slope) and projected closing pressure P_close (horizontal zero-crossing) were calculated.

Data Analysis: All slices were grouped into the retropalatal and retroglossal regions. For the adolescent OSA patients, compliance and P_close of the inhale and exhale portion of the first two breaths within one occlusion were calculated and compared during sleep and wakefulness respectively. Airway collapsibility was also compared across different subject categories.

Figs. 1 & 2 contain some representative results from one OSA patient during sleep. Fig.1 shows two frames, one with the airway open (top row) and the other with it partially collapsed (bottom row). The proposed reconstruction was able to recover all of the relevant UA boundary information. Minor residual streaking artifacts persisted but did not affect airway segmentation in our experience. This could be mitigated by sacrificing temporal resolution but a ~100ms resolution was purposely chosen to fully resolve the airway dynamics. Fig. 2a shows the cross-sectional area of each slice together with the mask pressure. Fig. 2b shows the linear regression lines for all four slices. One important finding is that a narrower airway site during tidal breathing does not necessarily have higher compliance or P_close, and therefore is not always the most collapsible. Table 1 & 2 show that compliance and P_close during sleep had smaller variance among the inhale/exhale portions of different breaths when compared to wakefulness. This is likely due to the involuntary muscle tone change, which suggests only the inhale portion of the first occluded breath should be used if only a wakefulness scan is performed. Table 3 shows the difference of compliance and P_close values between OSA and healthy subjects was significant and both measures could potentially serve as biomarkers for diagnosing OSA.we have demonstrated a novel imaging method for airway collapsibility measurement that combines acceleration techniques including SMS, parallel imaging, modified GA radial trajectory, and compressed sensing, to achieve 33.3x acceleration compared to fully sampled Cartesian scanning. To our best knowledge, we have experimentally discovered for the first time that a narrower airway site does not always correspond to higher collapsibility. This finding may be of interest to sleep surgeons. Our preliminary results suggest that both compliance and P_close may serve as biomarkers to diagnose OSA, and can be calculated with a 20-second awake scan.