基于鼾声检测的睡眠呼吸暂停低通气综合症诊断

发布时间：2018-06-30 00:06

本文选题：熟声 + 呼吸暂停　；参考：《大连理工大学》2010年硕士论文

【摘要】： 睡眠呼吸暂停低通气综合征,是一种具有严重危害及潜在危险的高发性疾病。由于在睡眠时时常发生低氧及高碳酸血症,最终引起全身多系统、多器官的渐进性危害,患者的身体健康状况受到很大的影响。一直以来,多导睡眠图是公认的诊断睡眠呼吸暂停低通气综合症的“金标准”。但是多导睡眠图监测设备数量有限,绝大多数的患者得不到及时的诊治。一种低费用的,更加便捷的诊断方法得到众多研究者的青睐。本文研究基于鼾声检测的方法筛查睡眠呼吸暂停低通气综合症。首先基于语音处理的方法对鼾声信号预处理,其次在讨论了传统的短时能量、短时过门限率和双门限端点检测的基础上,针对鼾声信号的特性提出了自适应的短时能量端点检测法。与传统的固定阈值端点检测相比,新算法能够动态的调整能量阈值从而有效地检测出有声段。由于打鼾的不连续性,本文对鼾声事件做了合理的修正：将间隔时间小于0.5s的相邻两段鼾声合并为一段；将间隔时间大于60s排除不做考虑。本文提出了三种方法对睡眠呼吸暂停低通气综合症诊断： (1)定义鼾声间隔时间大于10s为呼吸暂停,若单位时间内呼吸暂停的次数大于5次则判断为有病,反之,判断为无病。 (2)根据男女不同性别之间的差异按照不同的性别分别诊断,男性病例以单位时间呼吸暂停次数等于8为判断阈值,女性病例以单位时间呼吸暂停次数等于6为判断阈值。 (3)为了进一步减小个体差异带来的影响,基于K均值对鼾声间隔时间分类为正常的间隔时间和呼吸暂停的间隔时间,根据两类中心的距离和呼吸暂停次数判断是否患病。实验结果表明,三种方法都能够实现对睡眠呼吸暂停低通气综合症的筛查,灵敏度和特异度分别高达100%和85.7%。
[Abstract]:Sleep apnea hypopnea syndrome (SPAHS) is a high risk disease with serious hazards and potential dangers. Hypoxia and hypercapnia often occur during sleep, which eventually lead to systemic multi-system, multi-organ progressive harm, and the patient's health is greatly affected. Polysomnography has long been recognized as the gold standard for the diagnosis of sleep apnea hypopnea syndrome. However, the number of polysomnography monitoring equipment is limited, and the majority of patients can not be treated in time. A low-cost, more convenient diagnostic method has been favored by many researchers. The aim of this study was to screen for sleep apnea hypopnea syndrome based on snoring detection. Firstly, the snoring signal is preprocessed based on speech processing. Secondly, the traditional short time energy, short time threshold rate and double threshold endpoint detection are discussed. An adaptive short-time energy endpoint detection method is proposed for snoring signal. Compared with the traditional fixed threshold endpoint detection, the new algorithm can dynamically adjust the energy threshold to detect the acoustic segment effectively. Due to the discontinuity of snoring, this paper makes a reasonable correction to the snoring event: merging the two adjacent snoring segments with interval time less than 0.5 s into one segment, and excluding the interval time greater than 60 s. This paper presents three methods for diagnosis of sleep apnea hypopnea syndrome: (1) define snoring interval longer than 10 s as apnea, and if the number of apnea per unit time is more than 5 times, it can be judged as diseased. On the other hand, it was judged as disease-free. (2) according to the difference of sex between men and women, the male cases were diagnosed according to different gender, and the threshold value of apnea per unit time was equal to 8. The threshold of apnea per unit time was equal to 6. (3) in order to further reduce the impact of individual differences, The snoring interval time was classified as normal interval time and apnea interval time based on K-means, and the distance between the two types of centers and the number of apnea were used to judge whether the disease was ill or not. The results show that the three methods can be used to screen sleep apnea hypopnea syndrome with sensitivity of 100% and specificity of 85.7%.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：R766

【引证文献】