Automatic Segmentation and Quantification of Upper Airway Anatomic Risk Factors for Obstructive Sleep Apnea on Unprocessed Magnetic Resonance Images

      Rationale and Objectives

      Accurate segmentation of the upper airway lumen and surrounding soft tissue anatomy, especially tongue fat, using magnetic resonance images is crucial for evaluating the role of anatomic risk factors in the pathogenesis of obstructive sleep apnea (OSA). We present a convolutional neural network to automatically segment and quantify upper airway structures that are known OSA risk factors from unprocessed magnetic resonance images.

      Materials and Methods

      Four datasets (n = [31, 35, 64, 76]) with T1-weighted scans and manually delineated labels of 10 regions of interest were used for model training and validations. We investigated a modified U-Net architecture that uses multiple convolution filter sizes to achieve multi-scale feature extraction. Validations included four-fold cross-validation and leave-study-out validations to measure generalization ability of the trained models. Automatic segmentations were also used to calculate the tongue fat ratio, a biomarker of OSA. Dice coefficient, Pearson's correlation, agreement analyses, and expert-derived clinical parameters were used to evaluate segmentations and tongue fat ratio values.


      Cross-validated mean Dice coefficient across all regions of interests and scans was 0.70 ± 0.10 with highest mean Dice coefficient in the tongue (0.89) and mandible (0.81). The accuracy was consistent across all four folds. Also, leave-study-out validations obtained comparable accuracy across uniquely acquired datasets. Segmented volumes and the derived tongue fat ratio values showed high correlation with manual measurements, with differences that were not statistically significant (p < 0.05).


      High accuracy of automated segmentations indicate translational potential of the proposed method to replace time consuming manual segmentation tasks in clinical settings and large-scale research studies.

      Key Words


      OSA (obstructive sleep apnea), MRI (magnetic resonance imaging), MR (magnetic resonance), DL (deep learning), CNN (convolutional neural network), ROI (region of interest), V1 (voxel-weighted at max fat pad intensity), V2 (voxel-weighted at 99th percentile fat pad intensity)
      To read this article in full you will need to make a payment

      Purchase one-time access:

      Academic & Personal: 24 hour online accessCorporate R&D Professionals: 24 hour online access
      One-time access price info
      • For academic or personal research use, select 'Academic and Personal'
      • For corporate R&D use, select 'Corporate R&D Professionals'


      Subscribe to Academic Radiology
      Already a print subscriber? Claim online access
      Already an online subscriber? Sign in
      Institutional Access: Sign in to ScienceDirect


        • Benjafield AV
        • Ayas NT
        • Eastwood PR
        • et al.
        Estimation of the global prevalence and burden of obstructive sleep apnoea: a literature-based analysis.
        Lancet Respir Med. 2019; 7: 687-698
        • Patil SP
        • Schneider H
        • Schwartz AR
        • et al.
        Adult obstructive sleep apnea.
        Chest. 2007; 132: 325-337
        • Schwab RJ
        • Pasirstein M
        • Pierson R
        • et al.
        Identification of upper airway anatomic risk factors for obstructive sleep apnea with volumetric magnetic resonance imaging.
        Am J Respir Crit Care Med. 2003; 168: 522-530
        • Wang SH
        • Keenan BT
        • Wiemken A
        • et al.
        Effect of weight loss on upper airway anatomy and the apnea–Hypopnea index. The importance of tongue fat.
        Am J Respir Crit Care Med. 2020; 201: 718-727
        • Kim AM
        • Keenan BT
        • Jackson N
        • et al.
        Tongue fat and its relationship to obstructive sleep apnea.
        Sleep. 2014; 37: 1639-1648
        • LeCun Y
        • Bengio Y
        • Hinton G.
        Deep learning.
        Nature. 2015; 521: 436-444
        • Kleesiek J
        • Urban G
        • Hubert A
        • et al.
        Deep MRI brain extraction: a 3D convolutional neural network for skull stripping.
        NeuroImage. 2016; 129: 460-469
        • Havaei M
        • Davy A
        • Warde-Farley D
        • et al.
        Brain tumor segmentation with deep neural networks.
        Med Image Anal. 2017; 35: 18-31
        • Wu W
        • Yu Y
        • Wang Q
        • et al.
        Upper airway segmentation based on the attention mechanism of weak feature regions.
        IEEE Access. 2021; 9: 95372-95381
        • Park J
        • Hwang JJ
        • Ryu J
        • et al.
        Deep learning based airway segmentation using key point prediction.
        Appl Sci. 2021; 11: 3501
        • Alsufyani NA
        • Flores-Mir C
        • Major PW.
        Three-dimensional segmentation of the upper airway using cone beam CT: a systematic review.
        Dentomaxillofac Radiol. 2012; 41: 276-284
        • Shahid ML
        • Chitiboi T
        • Ivanovska T
        • et al.
        Automatic MRI segmentation of para-pharyngeal fat pads using interactive visual feature space analysis for classification.
        BMC Med Imaging. 2017; 17 0179-7
        • Ivanovska T
        • Dober J
        • Laqua R
        • et al.
        Pharynx segmentation from MRI data for analysis of sleep related disorders.
        Adv Vis Comput. 2013; : 20-29
        • Xie L
        • Udupa JK
        • Tong Y
        • et al.
        Automatic upper airway segmentation in static and dynamic MRI via anatomy-guided Convolutional Neural Networks.
        Med Phys. 2021; 49: 324-342
      1. Doshi J, Erus G, Habes M, et al. DeepMRSeg: a convolutional deep neural network for anatomy and abnormality segmentation on MR images. arXiv. Preprint posted online July 3, 2019. Available at: Accessed 01 Mar 2022.

        • Ronneberger O
        • Fischer P
        • Brox T.
        U-Net: convolutional networks for biomedical image segmentation.
        Lect Notes Comput Sci. 2015; : 234-241
        • He K
        • Zhang X
        • Ren S
        • et al.
        Deep residual learning for image recognition.
        in: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016
      2. Szegedy C, Ioffe S, Vanhoucke V, et al. Inception-v4, inception-ResNet and the impact of residual connections on learning. arXiv. Preprint posted online February 23, 2016. Available at: Accessed 01 Mar 2022.

        • Humbert IA
        • Reeder SB
        • Porcaro EJ
        • et al.
        Simultaneous estimation of tongue volume and fat fraction using ideal-FSE.
        J Magn Reson Imaging. 2008; 28: 504-508
        • Martin Bland J
        • Altman DG
        Statistical methods for assessing agreement between two methods of clinical measurement.
        Lancet. 1986; 327: 307-310
        • Cohen J.
        Statistical power analysis for the behavioral sciences.
        Psychology Press, New York, NJ2009
        • McHugh ML.
        Interrater reliability: the kappa statistic.
        Biochem Med (Zagreb). 2012; 22: 276-282
        • Pro Schwab RJ.
        Sleep apnea is an anatomic disorder.
        Am J Respir Crit Care Med. 2003; 168: 270-271