ISSN:
1573-2614
Keywords:
chronic obstructive pulmonary disease
;
nocturnal hypoxaemia
;
sleep stages
;
sleep study
;
computer analysis
;
thoracoplasty
Source:
Springer Online Journal Archives 1860-2000
Topics:
Computer Science
,
Medicine
Notes:
Abstract A computer-assisted method for the evaluation of sleep and breathing in patients showing chronic ventilatory impairment is described and validated. Signals of body and respiratory movements (static charge sensitive bed), air-flow (thermistors), oxygen saturation (SaO2), electro-oculography (EOG), and electromyography (EMG) were recorded overnight and analysed. Using the compressed output graphs of the data and a rapid scoring procedure, stages of wakefulness, non-REM (stages S1–S4) and REM sleep were identified. The procedure allowed analysis of oxygen saturation data separately for each sleep stage. For validation of the method, the sleep stages identified were compared with traditional sleep staging based on a simultaneous recording of EEG, EMG and EOG in 10 patients with chronic obstructive pulmonary disease (COPD) and in 15 patients treated by thoracoplasty (TPL) for pulmonary tuberculosis. The recordings were performed in a patient ward. In total, 32 night recordings were analysed. In the COPD patients, the sensitivity and specificity of the new method were 87% and 84% in detecting non-REM sleep, and 72% and 87% in detecting REM sleep, respectively. In the TPL patients the sensitivity and specificity were 93% and 89% with respect to non-REM sleep, and 92% and 94% in regard to REM sleep. The new method and traditional sleep staging provided closely similar quantitative estimates of the degree of sleep stage-(REM and non-REM) dependent arterial oxygen desaturation. It is concluded that the computer-assisted method, which is considerably less time consuming than traditional polysomnography, is reliable in studying sleep-related oxygenation in patients with chronic lung diseases.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1007/BF01132846
Permalink
