Ferri Raffaele, Gagnon Jean-François, Postuma Ronald B, Rundo Francesco, Montplaisir Jacques Y
Sleep Research Centre, Department of Neurology I.C., Oasi Institute (IRCCS), Troina, Italy.
Centre d'Études Avancées en Médecine du Sommeil, Hôpital du Sacré-Coeur de Montréal, Québec, Canada; Department of Psychology, Université du Québec à Montréal, Montreal, Quebec, Canada.
Sleep Med. 2014 Jun;15(6):661-5. doi: 10.1016/j.sleep.2013.12.022. Epub 2014 Mar 5.
To compare two different methods, one visual and the other automatic, for the quantification of rapid eye movement (REM) sleep without atonia (RSWA) in the diagnosis of REM sleep behavior disorder (RBD).
Seventy-four RBD patients (mean age, 62.14±9.67 years) and 75 normal controls (mean age, 61.04±12.13 years) underwent one night video-polysomnographic recording. The chin electromyogram (EMG) during REM sleep was analyzed by means of a previously published visual method quantifying the percentage of 30s epochs scored as tonic (abnormal, > or =30%) and that of 2s mini-epochs containing phasic EMG events (abnormal, > or =15%). For the computer quantitative analysis we used the automatic scoring algorithm known as the atonia index (abnormal, <0.8). The percentage correct classification, sensitivity, specificity, and Cohen kappa were calculated.
The atonia index correctly classified 82.6% of subjects, similar to the percentage of correct classifications with individual components of the visual analysis (83.2% each for tonic and phasic), and the combined visual parameters (85.9%). The sensitivity and specificity of automatic analysis (84% and 81%) was similar to the combined visual analysis (89% and 83%). The correlation coefficient between the automatic atonia index and the percentage of visual tonic EMG was high (r = -0.886, P<0.00001), with moderately high correlation with the percentage of phasic EMG (r = -0.690, P<0.00001). The agreement between atonia index and the visual parameters (individual or combined) was approximately 85% with Cohen's kappa, ranging from 0.638 to 0.693.
Sensitivity, specificity, and correct classifications were high with both methods. Moreover, there was general agreement between methods, with Cohen's kappa values in the 'good' range. Given the considerable practical advantages of automatic quantification of REM atonia, automatic quantification may be a useful alternative to visual scoring methods in otherwise uncomplicated polysomnograms.
比较两种不同方法（一种是视觉分析，另一种是自动分析）在快速眼动（REM）睡眠无张力（RSWA）定量分析中对快速眼动睡眠行为障碍（RBD）的诊断价值。
74例RBD患者（平均年龄62.14±9.67岁）和75例正常对照者（平均年龄61.04±12.13岁）接受了一晚的视频多导睡眠图记录。通过一种先前发表的视觉方法分析REM睡眠期间的颏肌肌电图（EMG），该方法量化了30秒时段被评为紧张性（异常，≥30%）的百分比以及包含相位性EMG事件的2秒微时段的百分比（异常，≥15%）。对于计算机定量分析，我们使用了称为无张力指数的自动评分算法（异常，<0.8）。计算正确分类百分比、敏感性、特异性和科恩kappa值。
无张力指数正确分类了82.6%的受试者，与视觉分析单个成分的正确分类百分比（紧张性和相位性各为83.2%）以及综合视觉参数（85.9%）相似。自动分析的敏感性和特异性（84%和81%）与综合视觉分析（89%和83%）相似。自动无张力指数与视觉紧张性EMG百分比之间的相关系数很高（r = -0.886，P<0.00001），与相位性EMG百分比的相关性中等偏高（r = -0.690，P<0.00001）。无张力指数与视觉参数（单个或综合）之间的一致性约为85%，科恩kappa值范围为0.638至0.693。
两种方法的敏感性、特异性和正确分类率都很高。此外，两种方法之间总体一致，科恩kappa值处于“良好”范围。鉴于REM无张力自动定量分析具有相当大的实际优势，在其他方面不复杂的多导睡眠图中，自动定量分析可能是视觉评分方法的一种有用替代方法。
