An automatic spindle analysis and detection system based on the evaluation of human ratings of the spindle quality

Abstract

An evaluation of the spindle quality ranked by experienced human raters is described. The pattern discriminating criteria obtained from this evaluation are used for an automatic system based on the complex demodulation method. The performance of this automatic system is compared with both the human visual scoring and with a second automatic system employing phaselocked loop techniques. The performance of the complex demodulation is satisfactory. It also detects spindles buried in slow waves. It is not sensitive to the age of the subject. This method opens possibilities for research of either temporal or chemical (drugs etc.) effects on the characteristics of rhythmic activity, without being hampered by the arbitrary EEG jargon "alpha" or "sigma".