阻塞性睡眠呼吸中止症（OSA）是一種慢性呼吸障礙，上呼吸道反覆塌陷導致睡眠期間氣流停止。米勒（Müller）動作是一種用於模擬清醒期間上咽部氣道塌陷的技術。要執行此技術，個人應在鼻子輕微捏住的同時閉上嘴巴試圖吸氣。米勒（MM）動作誘發的負壓導致上咽部氣道（UA）塌陷。舌頭因過度活動而表現出顯著的多方向位移，在進行MM過程中尤其具有挑戰性的運動追蹤。先前的研究提出了各種舌頭運動追蹤方法，每種方法都有其局限性。修改的光流方法（MOFM）通過結合不同技術來應對這一挑戰，捕捉視頻序列中連續幀之間的時間動態並估計像素位移。已使用不同的成像模式直接可視化UA解剖結構並定位阻塞點。本研究利用US的潛在優勢預測OSA發生，因為這在先前的研究中已被使用，同時檢測UA旁軟組織的動態變化，包括舌頭，有無MM的情況下。動態舌運動（DTM）的定量評估可以提供有關OSA發病機制的關鍵信息，有助於其診斷。使用MOFM跟踪在清醒狀態下由米勒（Müller）動作引起的下巴下US圖像中的舌頭面積變化。識別舌頭的不同模式和變形以及位移可能在評估OSA的存在和發病機制方面具有潛力。該研究的初步結果顯示，OSA患者的最大舌頭厚度（mTT）明顯較大（p = 0.005），無論有無進行MM（p = 0.004）。此外，舌面積測量結果顯示，在NB期間（p = 0.005）和進行MM後（p = 0.008），四組之間存在明顯差異。最近的努力利用人工智能技術來分類OSA嚴重程度，利用心電圖和血氧飽和度數據。儘管如此，對舌頭的US成像在發展旨在確定OSA嚴重程度的機器學習模型方面仍然被大多數忽視。本研究試圖通過捕捉清醒狀態下的DTM動態，包括從NB到進行MM的過渡，來彌合這一差距。由於其能夠減輕不平衡數據集中的偏見，系統評估選擇了平均準確率（MA），而不是混淆矩陣。對於兩級OSA數據集分類，取得了令人稱道的平均準確率（MA）為70.53％。此外，對於三級OSA嚴重程度分類（正常、輕度OSA和嚴重OSA），得到了57.81％的MA。提出的人工智能（AI）模型可以識別不同的OSA嚴重程度水平。

Obstructive sleep apnea (OSA) is a chronic breathing disorder where the recursive collapse of the upper airway causes cessation of airflow during sleep. Müller maneuver is a technique used to simulate the collapse of the upper pharyngeal airway during wakefulness. To perform this technique, an individual shall attempt to inhale while slightly pinching their nasal while their mouth is closed. The negative pressure induced by the Müller maneuver (MM) causes the upper pharyngeal airway (UA) to collapse. The tongue, being hypermobile, exhibits significant multidirectional displacements, posing challenges in motion tracking, especially during the MM performance. Previous studies proposed various tongue motion tracking methods, each with its limitations. The modified optical flow method (MOFM) addresses this challenge by combining different techniques, capturing temporal dynamics and estimating pixel displacements between consecutive frames in video sequences. Different imaging modalities have been used to visualize the UA anatomy directly and to localize the sites of obstruction. The potential advantages of US in predicting the occurrence of OSA have been utilized in this study because this has been used in previous studies, also in detecting dynamic variations of the soft tissue adjacent to the UA, the tongue, with and without the performance of MM. Quantitative assessments of dynamic tongue movement (DTM) can provide key information regarding the pathophysiological mechanism underlying OSA, facilitating its diagnosis. Tongue area changes, in submental US images recorded during wakefulness from normal breathing (NB) by Müller maneuver, were tracked using the MOFM. Identifying different patterns and deformation and displacement of the tongue may have potential in evaluating the presence and the pathogenesis of OSA. The initial result from this study demonstrated a significant larger maximum tongue thickness (mTT) in patients with OSA with (p=.005) and without the MM (p=.004). Additionally, tongue area measurements indicated a notably difference among the four groups during NB (p=.005) and after the performance of the MM (p=.008). Recent endeavors have utilized artificial intelligence techniques to categorize OSA severity leveraging electrocardiography and blood oxygen saturation data. Nonetheless, the integration of US imaging of the tongue remains largely untapped in the development of machine learning models aimed to determining the severity of OSA. This study endeavors to bridge this gap by capturing US images of DTM dynamics during wakefulness, encompassing transitions from NB to the performance of the MM. Mean accuracy (MA), instead of the confusion matrix, was chosen for system evaluation due to its ability to mitigate bias in an imbalanced dataset. A commendable mean accuracy (MA) of 70.53% was achieved for the two-level OSA dataset classification. Additionally, an MA of 57.81% was yielded for the three-level OSA severity classification (normal, mild OSA, and severe OSA). The proposed artificial intelligence (AI) model discerns varying OSA severity levels.

Al-hammuri K, Gebali F, Thirumarai Chelvan I, Kanan A. Tongue Contour Tracking and Segmentation in Lingual Ultrasound for Speech Recognition: A Review. Diagnostics 2022; 11.
Al-Shawwa B, Badi A, Goldberg A, Woodson B. Defining common outcome metrics used in obstructive sleep apnea. Sleep Med Rev 2008:449-4461.
Amiel Nico G. Loceo, Robert Brandan Lim, Pellegrino RV. 2023 Monitoring of Breathing Effort and Oxygen Levels for Identification of Sleep Apnea. International Conference on Computer and Automation Engineering (ICCAE): IEEE.
Appleton S, Vakulin A, McEvoy R, Vincent A, Martin S, Grant J, Taylor A, Antic N, Catcheside P, Wittert G, Adams R. Undiagnosed obstructive sleep apnea is independently associated with reductions in quality of life in middle-aged, but not elderly men of a population cohort. Sleep Breath 2015; 19:1309-16.
Barrera JE. Sleep magnetic resonance imaging: dynamic characteristics of the airway during sleep in obstructive sleep apnea syndrome. Laryngoscope 2011; 121:1327-35.
Bengio Y, Simard P, Frasconi P. Learning Long-Term Dependencies with Gradient Descent is Difficult. IEEE Transactions on Neural Networks 1994; 5:157-66.
Bernardini A, Brunello A, Gigli GL, Montanari A, Saccomanno N. AIOSA: An approach to the automatic identification of obstructive sleep apnea events based on deep learning. Artif Intell Med 2021; 118:102133.
Biswal S, Sun J, Sun H, Westover MB, Goparaju B, Bianchi MT. Expert-level sleep scoring with deep neural networks. Journal of the American Medical Informatics Association 2018; 25:1643-50.
Brown EC, Cheng S, McKenzie DK, Butler JE, Gandevia SC, Bilston LE. Respiratory Movement of Upper Airway Tissue in Obstructive Sleep Apnea. Sleep 2013; 36:1069-76.
Chen JW, Chang CH, Wang SJ, Chang YT, Huang CC. Submental ultrasound measurement of dynamic tongue base thickness in patients with obstructive sleep apnea. Ultrasound Med Biol 2014; 40:2590-8.
Chien C, Chen J, Chang C, Huang C. Tracking Dynamic Tongue Motion in Ultrasound Images for Obstructive Sleep Apnea. Ultrasound Med Biol 2017; 43:2791-805.
Cho K, Merriënboer Bv, Bahdanau D, Bengio Y. On the Properties of Neural Machine Translation: Encoder-Decoder Approaches. 2014:103-11.
Cho K, Merriënboer BV, Gulcehre C, Bahdanau D, Bougares F, Schwenk H, Bengio Y. Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation. 2014:1724–34.
Chousangsuntorn K, Bhongmakapat T, Apirakkittikul N, Sungkarat W, Supakul N, Laothamatas J. CT Characterization and Comparison with Polysomnography for Obstructive Sleep Apnea Evaluation. J Oral Maxillofac Surg 2018; 76:854-72.
Chu CA, Chen YJ, Chang KV, Wu WT, Ozcakar L. Reliability of Sonoelastography Measurement of Tongue Muscles and Its Application on Obstructive Sleep Apnea. Front Physiol 2021; 12:654667.
Chung J, Gulcehre C, Cho K, Bengio Y. Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling. arXiv 2014.
Ciscar MA, Juan G, Martinez V, Ramon M, Lloret T, Minguez J, Armengot M, Marin J, Basterra J. Magnetic resonance imaging of the pharynx in OSA patients and healthy subjects. Eur Respir J 2001; 17:79-86.
Cori JM, O'Donoghue FJ, Jordan AS. Sleeping tongue: current perspectives of genioglossus control in healthy individuals and patients with obstructive sleep apnea. Nat Sci Sleep 2018; 10:169-79.
Dzmitry B, Cho K, Bengio Y. Neural Machine Translation by Jointly Learning to Align and Translate. 2014.
Epstein LJ, Kristo D, Strollo PJJ, Friedman N, Malhotra A, Patil SP, Ramar K, Rogers R, Schwab RJ, Weaver EM, Weinstein MD. Clinical Guideline for the Evaluation, Management and Long-term Care of Obstructive Sleep Apnea in Adults. J Clin Sleep Med 2009; 5:263-76.
Faber CE, Grymer L. Available techniques for objective assessment of upper airway narrowing in snoring and sleep apnea. Sleep Breath 2003; 7:77-86.
Fogel RB, Malhotra A, Pillar G, Edwards JK, Beauregard J, Shea SA, White DP. Genioglossal activation in patients with obstructive sleep apnea versus control subjects. Mechanisms of muscle control. Am J Respir Crit Care Med 2001; 164:2025-30.
Gao S, Zheng Y, Guo X. Gated recurrent unit-based heart sound analysis for heart failure screening. Biomed Eng Online 2020; 19:3.
Hochreiter S, Schmidhuber J. Long Short Term Memory. Neural Computation 1997; 9:1735-80.
Hong SN, Won TB, Kim JW, Lee CH, Rhee CS. Upper Airway Evaluation in Patients with Obstructive Sleep Apnea. Sleep Med Res 2016; 7:1-9.
Hudgel DW. Sleep Apnea Severity Classification - Revisited. Sleep 2016; 39:1165-6.
Ikeda K, Ogura M, Oshima T, Suzuki H, Higano S, Takahashi S, Kurosawa H, Hida W, Matsuoka H, Takasaka T. Quantitative assessment of the pharyngeal airway by dynamic magnetic resonance imaging in obstructive sleep apnea syndrome. Ann Otol Rhinol Laryngol 2001; 110:183-9.
Isaiah A, Mezrich R, Wolf J. Ultrasonographic Detection of Airway Obstruction in a Model of Obstructive Sleep Apnea. Ultrasound Int Open 2017; 3:E34-E42.
Jeng-Wen Chen S-TL, Cheng-Yi Wang, Chun-Cheng Lin, Kuan-Chun Hsu, Cheng-Yu Yeh, Shaw-Hwa Hwang. A deep neural network-based model for OSA severity classification using unsegmented peripheral oxygen saturation signals. Engineering Applications of Artiicial Intelligence 2023; 122.
Jeng-Wen Chen S-TL, Cheng-Yi Wang, Chun-Cheng Lin, Kuan-Chun Hsu, Cheng-Yu Yeh, Shaw-Hwa Hwang. A Signal Segmentation-Free Model for Electrocardiogram-Based Obstructive Sleep Apnea Severity Classification. Advanced Intelligent Systems 2023; 5.
Jiawei Zhang, Jinshan Pan, Jimmy Ren, Yibing Song, Linchao Bao, Rynson W.H. Lau, Yang M-H. 2018 Dynamic Scene Deblurring Using Spatially Variant Recurrent Neural Networks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, USA: IEEE.
Joe Yue-Hei Ng, Matthew Hausknecht, Sudheendra Vijayanarasimhan, Oriol Vinyals, Rajat Monga, Toderici G. 2015 Beyond short snippets: Deep networks for video classification. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, MA, USA: IEEE Computer Society, 4694-702.
John A, Nundy KK, Cardiff B, John D. Multimodal Multiresolution Data Fusion Using Convolutional Neural Networks for IoT Wearable Sensing. IEEE Trans Biomed Circuits Syst 2021; 15:1161-73.
Kapur V, Strohl KP, Redline S, Iber C, O'Connor G, Nieto J. Underdiagnosis of sleep apnea syndrome in U.S. communities. Sleep Breath 2002; 6:49-54.
Kim AM, Keenan BT, Jackson N, Chan EL, Staley B, Poptani H, Torigian DA, Pack AI, Schwab RJ. Tongue fat and its relationship to obstructive sleep apnea. Sleep 2014; 37:1639-48.
Kwan BC, Butler JE, Hudson AL, McKenzie DK, Bilston LE, Gandevia SC. A novel ultrasound technique to measure genioglossus movement in vivo. J Appl Physiol (1985) 2014; 117:556-62.
Kwon S, Kim H, Yeo WH. Recent advances in wearable sensors and portable electronics for sleep monitoring. iScience 2021; 24:102461.
Lahav Y, Rosenzweig E, Heyman Z, Doljansky J, Green A, Dagan Y. Tongue base ultrasound: a diagnostic tool for predicting obstructive sleep apnea. Ann Otol Rhinol Laryngol 2009; 118:179-84.
Li H, Lo Y, Wang C, Hsin L, Lin W, Fang T, Lee L. Dynamic Drug-Induced Sleep Computed Tomography in Adults with Obstructive Sleep Apnea. Sci Rep 2016; 6.
Liao LJ, Cho TY, Cheng PW, Wang CT, Lo WC, Huang TW. Submental Ultrasonography in Diagnosing Severe Obstructive Sleep Apnea Syndrome. J Med Ultrasound 2016; 24:107-11.
Lulich; SM, Berkson; KH, Jong Kd. Acquiring and visualizing 3D/4D ultrasound recordings of tongue motion. Journal of Phonetics 2018; 71:410-24.
Manlises CO, Chen JW, Huang CC. Dynamic tongue area measurements in ultrasound images for adults with obstructive sleep apnea. J Sleep Res 2020; 29:e13032.
Martens J, Sutskever I. Learning recurrent neural networks with hessian-free optimization. 2011:1033–40.
Myers KA, Mrkobrada M, Simel DL. Does this Patient have Obstructive Sleep Apnea? The Rational Clinical Examination Systematic Review. N Engl J Med 2013; 310:731-41.
Osman AM, Carter SG, Carberry JC, Eckert DJ. Obstructive sleep apnea: current perspectives. Nat Sci Sleep 2018; 10:21-34.
Paris Charilaou, Battat R. Machine learning models and over-fitting considerations. World J Gastroenterol 2022; 28:605-07.
Patel D, Upadhyay S. Optical Flow Measurement using Lucas kanade Method. IJCA 2013; 61:6-10.
Patil SP, Schneider H, Schwartz AR, Smith PL. Adult obstructive sleep apnea: pathophysiology and diagnosis. Chest 2007; 132:325-37.
Punjabi N. The epidemiology of adult obstructive sleep apnea. Proc Am Thorac Soc 2008; 5:136-43.
Remijn L, Weljers G, Sanden NN, Groen B, De Korte C. Ultrasound Imaging for Analyzing Lateral Tongue Movements during Mastication in Adults with Cerebral Palsy Compared with Adults without Oral Motor Disabilities. Ultrasound Med Biol 2015; 41:1784-93.
Remmers JE, Degroot WJ, Sauerland EK, Anch AM. Pathogenesis of Upper Airway Occlusion during Sleep. J Appl Physiol 1978; 44:931-38.
Saboisky JP, Stashuk DW, Hamilton-Wright A, Carusona AL, Campana LM, Trinder J, Eckert DJ, Jordan AS, McSharry DG, White DP, Nandedkar S, David WS, Malhotra A. Neurogenic changes in the upper airway of patients with obstructive sleep apnea. Am J Respir Crit Care Med 2012; 185:322-9.
Sandeli Priyanwada Kasthuri Arachchi , Timothy K. Shih, Hakim aNL. Modelling a Spatial-Motion Deep Learning Framework to Classify Dynamic Patterns of Videos. Applied Sciences 2020; 10.
Schoenberg SO, Floemer F, Kroeger H, Hoffmann A, Bock M, Knopp MV. Combined assessment of obstructive sleep apnea syndrome with dynamic MRI and parallel EEG registration: initial results. Invest Radiol 2000; 35:267-76.
Sharma S, Sharma S, Athaiya A. ACTIVATION FUNCTIONS IN NEURAL NETWORKS. International Journal of Engineering Applied Sciences and Technology (IJEAST) 2020; 4:310-16.
Shu CC, Lee P, Lin JW, Huang CT, Chang YC, Yu CJ, Wang HC. The use of sub-mental ultrasonography for identifying patients with severe obstructive sleep apnea. PLoS One 2013; 8:e62848.
Shudong Yang, Xueying Yu, Zhou Y. 2020 LSTM and GRU Neural Network Performance Comparison Study: Taking Yelp Review Dataset as an Example. 2020 International Workshop on Electronic Communication and Artificial Intelligence (IWECAI). Shanghai, China: IEEE.
Sloun RJGV, Cohen R, Eldar YC. Deep learning in Ultrasound Imaging. 2020; 108.
Soares M, Sallum A, Gonçalves M, Haddad F, Gregório L. Use of Muller's maneuver in the evaluation of patients with sleep apnea--literature review. Braz J Otorhinolaryngol 2009; 75:463-66.
Stuck B, Maurer J. Airway evaluation in obstructive sleep apnea. Sleep Med Rev 2008; 12:411-36.
Szeghalmy S, Fazekas A. A Comparative Study of the Use of Stratified Cross-Validation and Distribution-Balanced Stratified Cross-Validation in Imbalanced Learning. Sensors (Basel) 2023; 23.
Togeiro SM, Chaves CM, Jr., Palombini L, Tufik S, Hora F, Nery LE. Evaluation of the upper airway in obstructive sleep apnoea. Indian J Med Res 2010; 131:230-5.
Wei F, Yupeng C, Qiongying X. Survey on Reseach of RNN-Based Spatio-Temporal Sequences Prediction Algorithms. Journal on Big Data 2021; 3:97-110.
Wen W, Xu D, Xia Y. A novel traffic optimization method using GRU based deep neural network for the IoV system. PeerJ Comput Sci 2023; 9.
Weng CK, Chen JW, Lee PY, Huang CC. Implementation of a Wearable Ultrasound Device for the Overnight Monitoring of Tongue Base Deformation during Obstructive Sleep Apnea Events. Ultrasound Med Biol 2017; 43:1639-50.
White DP. Sleep apnea. Proc Am Thorac Soc 2006; 3:124-8.
Xu C, Brennick M, Wootton D. Image-based three-dimensional finite element modeling approach for upper airway mechanics. Conf Proc IEEE Eng Med Biol Soc 2005; 2005:2587-90.
Xu K, Yang Y, Stone M, Jaumard-Hakoun A, Leboullenger C, Dreyfus G, Roussel P, Denby B. Robust contour tracking in ultrasound tongue image sequences. Clin Linguist Phon 2016; 30:313-27.
Ying X. An Overview of Overfitting and its Solutions. Journal of Physics 2019; 1168.
Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med 1993; 328:1230-35.
Yucel A, Unlu M, Haktanir A, Acar M, Fidan F. Evaluation of the upper airway cross-sectional area changes in different degrees of severity of obstructive sleep apnea syndrome: cephalometric and dynamic CT study. AJNR Am J Neuroradiol 2005; 26:2624-9.
Zuxuan Wu XW, Yu-Gang Jiang, Hao Ye, Xue X. 2015 Modeling Spatial-Temporal Clues in a Hybrid Deep Learning Framework for Video Classification. MM '15: ACM Multimedia Conference. Brisbane, Australia: Association for Computing MachineryNew YorkNYUnited States, 461-70.