Software design of high-frequency oscillation detection and seizure onset zones localization system for epilepsy
Objective High-frequency oscillations (HFOs) are regarded as promising biomarkers of epileptic seizure onset zones ( SOZs) . Accurate detection of HFOs is critical for the preoperative localization of SOZs. However,the detection and localization methods are usually relatively complex,and specialized software that packages them and provides an intuitive data analysis interface is needed for the researchers and clinicians. Thus,we design an intuitive HFO detection and SOZ localization system software HFOLab in this paper. Methods By using a modular design technique,we design the HFO detection method and the SOZ localization method in two submodules,and visualize the detection results and feed them into the localization submodule. Considering the false discovery problem of the detection method and the demand for manual detection,we also design two sub-modules of auxiliary verification and auxiliary detection of HFOs to provide intuitive references for experts to verify and mark of HFOs,respectively. Results Initial testing of HFOLab is performed by using EEG data from 20 epilepsy patients. The results show that HFOLab features an intuitive data analysis interface to visualize the detection results,and provides intuitive references for expert verification and marking of HFOs. The average sensitivity ( Ripples:92. 4％;Fast ripples:90. 3％) and average false discovery rate ( Ripples:9. 2％;Fast ripples:10. 7％) for the detection of HFOs are satisfactory,and the overall sensitivity and specificity for the localization of SOZs are 78. 9％ and 77. 4％,respectively. Conclusions HFOLab is effective in detecting HFOs,localizing SOZs, and facilitating experts to verify and mark HFOs, and thus provides a new tool for clinical trials applying HFOs to localize SOZs.
万方数据
维普
