Correlation Histogram to Improve Classification of Acute Encephalopathy in Infants Using EEG

We extract pairs of EEG waveforms from short time intervals and represent the distribution of their correlation coefficients as a histogram. We investigate using this feature to predict, at an early stage, whether acute encephalopathy in infants will lead to severe sequelae.

Acute encephalopathy in infants can leave severe lasting effects. This study explores whether these effects can be predicted early by using an electroencephalogram, or EEG, which records the brain's electrical activity as waves.

We focus on how similar pairs of EEG waves are. Guided by knowledge used in diagnosis, we randomly select pairs of waves from short time intervals and calculate a similarity value for each pair. We then count how often different ranges of similarity occur. The resulting chart, called a correlation histogram, summarizes the relationships among many pairs of waves rather than relying on just one pair.

In tests using data from patients with acute encephalopathy, this feature gave better classification performance than the conventional method used for comparison. The difference was statistically significant. These results concern classification of the collected patient data; the study investigates information in EEG signals that may help with early prediction.

• Takahiro Kondo, Masashi Nishiyama, Yoshio Iwai, Masayoshi Oguri, Yoshihiro Maegaki,
  Band Correlation Histogram to Improve Classification of Acute Encephalopathy in Infants, Proceedings of IEEE 1st Global Conference on Life Sciences and Technologies (LifeTech), pp. 268 - 272, March 2019.

• 池本 龍矢, 大栗 聖由, 西山 正志, 前垣 義弘,
  けいれん重積型急性脳症と熱性けいれん重積の判別におけるエポック自動選択の検討,
  2026年度（第77回）電気・情報関連学会中国支部連合大会, R26-21-01, pp. 1 - 2, October 2026.
• 西山 正志, 臼井 愛美, 岩井 儀雄, 大栗 聖由, 前垣 義弘,
  相関ヒストグラムを用いた小児急性脳症の判別,
  電子情報通信学会論文誌 D, Vol.J99-D, No.12, pp.1132-1141, December 2016.