筋電波形を用いた睡眠時ブラキシズム自動検出手法の開発

摘要

睡眠時無呼吸症候群と睡眠時ブラキシズム（SB）が随伴する可能性を明らかにするため，睡眠時の長時間筋電図記録からSBの発生を自動検出するアルゴリズムの考案を行った．この方法では，まず，複数のSBのトレーニング波形から基ベクトルを構成する．これを用いて，対象波形の最適化直交展開を行い，第一主成分への集中度を求める．この集中度のピーク位置から，筋電図中のSBの波形を検出する．被験者8名の臨床データに対し，専門医の目視による判定と比較した結果，79．5％の感度，89．3％の特異度でSBを検出可能なことが確かめられた．%To investigate the relevance between the sleep apnea and the sleep bruxism (SB), we devised an algorithm to detect the occurrence of SB in the EMG of masseter muscle. EMG waveforms of SB were extracted based on the temporal change of the degree of concentration (DC) of the first expansion coefficient calculated in the optimal orthogonal transform. Clinical data of eight subjects were processed with this technique. The results were compared with those obtained by visual observation of medical specialists. The sensitivity and the specificity were 79.5% and 89.3% respectively, and the effectivenessrnof the proposed technique was verified