Deep learning-based automated segmentation of airway OCT images acquired during drug-induced sleep endoscopy (DISE)
Obstructive sleep apnea (OSA) is a common upper airway disorder with severe long-term health impacts. There is a critical need for methods that provide quantitative information on dynamic airway collapse during sleep; information not available from sleep studies or conventional Imaging (e.g., CT). We present an alternative minimally invasive approach that combines drug-induced sleep endoscopy (DISE) with optical coherence tomography (OCT) and deep learning. These studies generate a massive volume of data, hence automated methods are needed using deep learning/convolutional neural networks (CNNs) to segment images. This strategy enhances the identification of critical regions prone to collapse, providing a robust tool to optimize treatment for OSA.
Authors
- Ellen M. Hong (ORCID: https://orcid.org/0000-0002-0084-2924)
- Brian J. F. Wong (ORCID: https://orcid.org/0000-0001-6318-7384)
- Jessica R. Salas (ORCID: https://orcid.org/0000-0002-1585-1225)
- Dinh-Manh-Cuong Tran (ORCID: https://orcid.org/0009-0000-7161-1077)
- Danny Hung‐Chieh Chou (ORCID: https://orcid.org/0000-0001-9110-614X)
- Raksha Sreeramachandra Murthy (ORCID: https://orcid.org/0000-0002-5313-8068)
- Zhongping Chen (ORCID: https://orcid.org/0000-0002-4584-4560)
- Elizabeth Lee (ORCID: https://orcid.org/0000-0001-9143-3140)
- Zhikai Zhu (ORCID: https://orcid.org/0000-0002-1447-3170)
- Hyunmo Yang (ORCID: https://orcid.org/0009-0008-3112-8572)
- Hongyan Lei (ORCID: https://orcid.org/0009-0008-4152-6662)
Institutions
- Beckman Laser Institute and Medical Clinic (US)
- Ulsan National Institute of Science and Technology (KR)
Publication Details
- Published
- 2025-03-19
- DOI
- https://doi.org/10.1117/12.3043827
- Citations
- 1
- Primary Topic
- Obstructive Sleep Apnea Research
- Type
- article
- Field-Weighted Citation Impact
- 1.16