{"references": ["De Gennaro, L., Ferrara, M. Sleep spindles: an overview. Sleep Med\nRev; pp. 7:423-40, 2003.", "Ktonas, P.Y., Golemati, S., Xanthopoulos, P. , Sakkalis, V., Ortigueira,\nM.D, et al. Time-frequency analysis methods to quantify the timevarying\nmicrostructure of sleep EEG spindles: Possibility for dementia\nbiomarkers? J. of Neuroscience Methods, Vol 185-1: 133-142, 2009.", "Causa L., Held C.M., Causa J., Est\u00e9vez P.A., Perez C.A., Chamorro R.,\nGarrido M., Algar\u251c\u00a1n C., Peirano P. 2010. Automated sleep-spindle\ndetection in healthy children polysomnograms. s.l. : IEEE Trans Biomed\nEng.;57(9):2135-46, 2010.", "Steriade, M., Jones, E.G., Llinas, R.: Thalamic Oscillations and\nSignaling. Neuroscience Institute Publications. John Wiley & Sons, New\nYork (1990)", "Ahmed B., Redissi A., Tafreshi R. 2009. An automatic sleep spindle\ndetector based on wavelets and the teager energy operato. s.l. : Annual\nInternational Conference of the IEEE Engineering in Medicine and\nBiology Society. IEEE Engineering in Medicine and Biology Society.\nConference 1:2596-9, 2009.", "Duman, F., Erogul, O., Telatar, Z., & Yetkin, S. Automatic sleep\nspindle detection and localization algorithm. Antalya, Turkey, 2005.", "G\u00f6r\u251c\u255dr D., Halici U., Aydin H., Ongun G., Ozgen F., Leblebicioglu K.\n2003. , Sleep Spindles Detection Using Autoregressive Modeling. s.l. :\nProc. of ICANN/ICONIP, 2003.", "Ventouras E., Monoyiou E., Ktonas P., Paparrigopoulos T., Dikeos D.,\nUzunoglu N., Soldatos C. 2005. Sleep Spindle Detection Using Artificial\nNeural Networks Trained with Filtered Time-Domain EEG: A\nFeasibility Study. s.l. : Computer Methods and Programs in Biomedicine\n78(3):191-207, 2005.", "Duman F., Erdamar A., Erogul O., Telatar Z., Yetkin S. 2009. Efficient\nsleep spindle detection algorithm with decision tree. s.l. : Expert\nSystems with Applications, Vol. 36, No. 6. pp. 9980-9985, 2009.\n[10] Causa L., Held C.M., Causa J., Est\u00e9vez P.A., Perez C.A., Chamorro R.,\nGarrido M., Algar\u251c\u00a1n C., Peirano P. 2010. Automated sleep-spindle\ndetection in healthy children polysomnograms. s.l. : IEEE Trans Biomed\nEng.;57(9):2135-46, 2010.\n[11] Proakis, J., Manolakis, D., Digital Signal Processing, 4th Ed., Prentice-\nHall, 2006.\n[12] Omerhodzic, I., Avdakovic,S., Nuhanovic, A., Dizdarevic, K. and\nRotim, K. Energy Distribution of EEG Signal Components by Wavelet\nTransform, pp45-60 IInTech publishing, 2012\n[13] Rechtschaffen, A, Kales, A. A manual of standardised terminology,\ntechniques and scoring system for sleep stages of human subjects.\nWashington, DC: Public Health Service, U.S. Government Printing\nOffice; 1968.\n[14] Costa, J., Ortigueira, M., Batista, A. Short Time Fourier Transform and\nAutomatic Visual Scoring for the detection of Sleep Spindles. DOCEIS\n2012. Springer, IFIP AICT series v.372, p. 267-272.\n[15] Devuyst, S., Dutoit, T., Didier, J. F. et al. Automatic sleep spindle\ndetection in patients with sleep disorders. Conf. Proc. IEEE Eng. Med.\nBiol. Soc. 1: 3883-3886, 2006.\n[16] Costa, J., Ortigueira, M.D., Batista, A., Paiva, T., \"Threshold choice for\nautomatic spindle detection\". Proc. IWSSIP2012; 2012\n[17] Sch\u00f6nwald, S., Santa-Helena, E., Rossatto, R., Chaves, M. and Gerhardt,\nG. Benchmarking matching pursuit to find sleep spindles, Journal of\nNeuroscience Methods Vol 156 1-2: 314-321, 2006."]}

Sleep spindles are the most interesting hallmark of stage 2 sleep EEG. Their accurate identification in a polysomnographic signal is essential for sleep professionals to help them mark Stage 2 sleep. Sleep Spindles are also promising objective indicators for neurodegenerative disorders. Visual spindle scoring however is a tedious workload. In this paper three different approaches are used for the automatic detection of sleep spindles: Short Time Fourier Transform, Wavelet Transform and Wave Morphology for Spindle Detection. In order to improve the results, a combination of the three detectors is presented and comparison with human expert scorers is performed. The best performance is obtained with a combination of the three algorithms which resulted in a sensitivity and specificity of 94% when compared to human expert scorers.