ニューラルネットを用いた重心動揺の識別

書誌事項

タイトル別名
  • Learning and Evaluation of Stabilograms using Neural Network.

この論文をさがす

抄録

For focal diagnosis in patients complaining of vertigo and equilibrium disturbances, learning and evaluation of stabilograms using a neural network (NN) were conducted in healthy subjects and patients with labyrinthine and brain disturbances.<BR>Stabilograms were recorded with eyes open and closed for 60 seconds. Learning and evaluation of stabilograms were performed using a NN program developed by Anima corporation. The learning was practiced with stabilograms of healthy subjects and patients with lesions in the labyrinth and the brain. Using loaded NN formed by the learning, stabilograms of other healthy subjects and clinical cases were evaluated. Results of evaluation, such as generation ability of the loaded NN, were shown by a corresponding ratio. Correspondence showed that the lesions inferred from output values of the loaded NN corresponded to those indicated by clinical diagnosis.<BR>(1) On evaluation of stabilograms of healthy subjects and clinical cases, the cor-responding ratio was 75% in healthy subjects and 76% in clinical cases.<BR>(2) On evaluation of stabilograms of cases of labyrinthine and brain disturbances, the corresponding ratio was 72% and 67%, respectively.<BR>(3) On evaluation of stabilograms of healthy subjects and cases of labyrinthine and brain disturbances, the corresponding ratio was 60%, 46% and 60%, respectively.<BR>Learning and evaluation of stabilograms by the NN provide useful information for focal diagnosis in patients with vertigo and equilibrium disturbances.

収録刊行物

被引用文献 (8)*注記

もっと見る

参考文献 (14)*注記

もっと見る

詳細情報 詳細情報について

問題の指摘

ページトップへ