• Raffaele Ferri, Marco Zucconi, Mauro Manconi, Oliviero Bruni, Silvia Miano, Giuseppe Plazzi, and Luigi Ferini-Strambi.
• Sleep Research Centre, Department of Neurology I.C., Oasi Institute (IRCCS), Troina, Italy. rferri@oasi.en.it
• Sleep. 2005 Aug 1; 28 (8): 998-1004.

Study ObjectivesTo assess the performance of a new method for automatic detection of periodic leg movements during sleep.MethodsLeg movements during sleep were visually detected in the tibialis anterior muscles recordings of 15 patients with restless legs syndrome and 15 normal controls. Leg movements were detected automatically by means of a new computer method with which electromyogram signals are first digitally band-pass filtered and then rectified; subsequently, the detection of leg movements is performed by using 2 thresholds: one for the starting point and another to detect the end point of each leg movement. Sensitivity and false-positive rate were obtained; the American Sleep Disorders Association parameters were also computed, and the results analyzed by means of the Kendall W coefficient, the linear correlation coefficient and the Bland-Altman plots.SettingN/A.ParticipantsFifteen patients with restless legs syndrome and periodic leg movements and 15 controls.Measurements And ResultsHigh values of the Kendall W coefficient of concordance between automatic and visual analysis were found with values close to 1 and the linear correlation coefficient for leg movements index and total leg movements index was > 0.950 (p < .000001). The Bland-Altman plots provided the limits of agreement between visual and computer detection, which were -9.01 and +9.89 for the periodic leg movement index. None of the normal controls was found to have periodic leg movement indexes >5 after automatic analysis.ConclusionsOur method can be applied to the clinical evaluation of periodic leg movements during sleep, with some caution in patients with a low periodic leg movement indexes. Large-scale research application is possible and can be considered as reliable.

28 keywords...

hide…