Distinguishing Obstructive Sleep Apnea Using Electroencephalography Records

Authors : İlhan Umut, Hakan Üstünel, Güven Çentik, Erdem Uçar, Levent Öztürk

Pages : 622-634

Doi:10.35378/gujs.1229166

View : 151 | Download : 280

Publication Date : 2024-06-01

Article Type : Research Paper

Abstract :In this study, it was aimed to find out whether electroencephalographic (EEG) frequency bands can be used to distinguish people with obstructive sleep apnea (OSA) from those who do not have it. 11842 different cases taken from 121 patients suffering from OSA were combined with the case study of 30-person control group without sleep apnea. Apneas were highlighted at the respiration-record channels and EEG records which are concurrent with abnormal respiration cases were extracted from C4-A1 and C3-A2. Following that, they were examined with Fourier and Wavelet Transforms using a new software that was developed by us. The percentage values of Delta (0, 5-4 Hz), Theta (4-8 Hz), Alpha (8-13 Hz) and Beta (13-30 Hz) frequency bands were evaluated with the help of t-test and ROC Analysis to differentiate between apneas. The C3-A2 Beta (%) frequency level gave the highest distinguishing asset (AUC=0.662; p<0.001); however, the C3-A2 Alpha (%) level yielded the lowest distinguishing (AUC=0.536; p<0.001). Similarly, the C4-A1 Alpha (%) level produced the lowest distinguishing asset (AUC=0.536; p<0.001) whereas the C4-A1 Beta (%) frequency level gave the highest distinguishing asset (AUC=0.658; p<0.001). The chief finding of this study suggests that the EEG rates of patients with OSA differ from those of patients without OSA and following the changes at these channels may give rise to detection of apneas, and the Beta (%) yielded the most meaningful result among four different frequency bands in the study.
Keywords : Sleep apnea Digital signal processing Electroencephalography, Sleep apnea, Digital signal processing, Digital signal processing, Electroencephalography