CAJ | 학술논문

通过癫痫发作间期高频颅内脑电信号的记录和分析，实现一种基于概率模型的癫痫病灶定位自动算法。方法以一段时间内颅内脑电高频能量的整体波动性水平为指标，建立大数据概率模型以判断某电极是否覆盖致痫灶。结果本文分析了来自12例癫痫患者948个颅内电极的癫痫发作间期颅内脑电数据，与医生人工定位结果作对比，平均敏感性80.4％±17.3％，特异性87.7％±17.2％；模型的稳定性和性能随着数据量增加而提高。结论本文所提出高频能量波动性算法基于概率模型，不依赖个体化参数、自动化程度高、性能好，有良好的临床应用前景。
통과전간발작간기고빈로내뇌전신호적기록화분석，실현일충기우개솔모형적전간병조정위자동산법。방법이일단시간내로내뇌전고빈능량적정체파동성수평위지표，건립대수거개솔모형이판단모전겁시부복개치간조。결과본문분석료래자12례전간환자948개로내전겁적전간발작간기로내뇌전수거，여의생인공정위결과작대비，평균민감성80.4％±17.3％，특이성87.7％±17.2％；모형적은정성화성능수착수거량증가이제고。결론본문소제출고빈능량파동성산법기우개솔모형，불의뢰개체화삼수、자동화정도고、성능호，유량호적림상응용전경。
Objective To propose and implement an automatic method based on a probability model using intracranially recorded high-frequency brain activity for the identification of epileptogenic zone.Methods Big data probability model was constructed based on the characteristics of the epileptogenic activities described by the degree of overall high-frequency power fluctuation in order to identify whether a channel was located in epileptogenic zone or not.Results By using the probability model with 948-electrode data from 1 2 patients,and compared with the results marked by neurologists,80.4% ±1 7.3% sensitivity and 87.7% ±1 7.2%specificity were achieved. Conclusions Based on probability model,the proposed method does not rely on individual parameters,and possesses high degree of automation,good performance and a bright perspective in clinical diagnosis.