ABSTRACT
Since the 1970s, various automatic sleep spindles procedures have been implemented and presented in the literature. Unfortunately, their results are not easily comparable because the databases, the assessment methods and the terminologies employed are often radically different. In this study, we propose a systematic assessment method for any automatic sleep spindles detection algorithm. We apply this assessment method to our own automatic detection process in order to illustrate and legitimate its use. We obtain a global sensitivity of 70.20%, for a false positive proportion (relative to the total number of visually scored sleep spindles) of only 26.44% (False positive rate = 1.38% and specificity = 98.62%).
Subject(s)
Algorithms , Brain/physiopathology , Diagnosis, Computer-Assisted/methods , Electroencephalography/methods , Pattern Recognition, Automated/methods , Polysomnography/methods , Sleep Wake Disorders/diagnosis , Sleep Wake Disorders/physiopathology , Sleep , Adult , Female , Humans , Male , Middle Aged , Reproducibility of Results , Sensitivity and SpecificityABSTRACT
In this paper, we present an automatic method for K-complexes detection based on features extraction and the use of fuzzy thresholds. The validity of our process was examined on the basis of two visual K-complexes scorings performed on 5 excerpts of 30 minutes. Results were investigated through all different sleep stages. The algorithm provides global true positive rates of 61.72% and 60.94%, respectively with scorer 1 and scorer 2. The false positive proportions (compared to the total number of visually scored K-complexes) are of 19.62% and 181.25%, while the false positive rates estimated on a one 1 second resolution are only of 0.53% and 1.53%. These results suggest that our approach is completely suitable since its performances are similar to those of the human scorers.
Subject(s)
Algorithms , Diagnosis, Computer-Assisted/methods , Electroencephalography/methods , Pattern Recognition, Automated/methods , Sleep/physiology , Differential Threshold/physiology , Humans , Reproducibility of Results , Sensitivity and SpecificityABSTRACT
In this paper, we introduce a new automatic method for electrocardiogram (ECG) artifact elimination from the electroencephalogram (EEG) or the electrooculogram (EOG). It is based on a modification of the independent component analysis (ICA) algorithm which gives promising results while only using a single-channel EEG (or EOG) and the ECG. To check the effectiveness of our approach, we compared its correction rate with those obtained by ensemble average subtraction (EAS) and adaptive filtering (AF). For this purpose, we applied these algorithms to 10 excerpts of polysomnographic sleep recordings containing ECG artifacts and other typical artifacts (e.g. movement, sweat, respiration, etc.). Two hundred successive interference peaks were examined in each excerpt to compute correction rates. We found that our modified ICA was the most robust to various waveforms of cardiac interference and to the presence of others artifacts, with a correction rate of 91.0%, against 83.5% for EAS and 83.1% for AF.
Subject(s)
Algorithms , Artifacts , Diagnosis, Computer-Assisted/methods , Electrocardiography/methods , Electroencephalography/methods , Pattern Recognition, Automated/methods , Sleep Wake Disorders/diagnosis , Adult , Aged , Artificial Intelligence , Female , Humans , Male , Middle Aged , Principal Component Analysis , Reproducibility of Results , Sensitivity and Specificity , Signal Processing, Computer-AssistedABSTRACT
In this paper, we present a new automatic method for sleep spindle detection. It consist of a generalisation of the Schimicek's method that takes more types of artefacts into account and uses variable thresholds regarding the statistical properties of the signal. Validity of our process is examined on the basis of visual spindle scoring performed by an expert. Results obtained are compared to those obtained by Schimicek's method. For a specificity of 90%, we obtain a sensitivity of 76.9% while Schimicek's method has a sensitivity of 70.4%. Moreover an increase of the area under the ROC curve is observed and confirms that the detection process is improved.
Subject(s)
Sleep Wake Disorders/diagnosis , Sleep/physiology , Algorithms , Automation , Electroencephalography , Electrooculography , False Positive Reactions , Humans , Polysomnography , Sensitivity and Specificity , Sleep Stages , WakefulnessABSTRACT
BACKGROUND: This study examines the operating characteristics of the expiratory flow response to a negative pressure (NEP) applied to the mouth in the prediction of obstructive sleep apnoea syndrome (OSAS) in snoring patients. METHODS: Two hundred and thirty eight patients with normal spirometric values were studied. Full laboratory polysomnography was performed and an NEP of -5 cm H(2)O was applied in the sitting and supine positions. RESULTS: A significant correlation was found between the degree of flow limitation measured by NEP in both positions (expressed as the percentage of the expired tidal volume over which NEP induced flow did not exceed spontaneous flow) and the apnoea-hypopnoea index (AHI). This correlation was significantly higher in the supine position (p<0.0001) where an expiratory flow limitation cut off value of >/=27.5% of the tidal volume produced a sensitivity of 81.9% and a specificity of 69.1% in predicting OSAS. CONCLUSION: These findings show that the degree of instability of the upper airway measured by NEP is correlated with the severity of OSAS. NEP had moderate sensitivity and specificity and may be useful in predicting OSAS in a clinic based population.
