基于机器学习的癫痫诊疗研究综述

REVIEW OF MACHINE LEARNING⁃BASED RESEARCH ON EPILEPSY DIAGNOSIS AND TREATMENT

癫痫是一种常见的神经系统疾病,其反复发作的病症严重影响着患者的身体和生活。随着人工智能的发展,机器学习(machine learning,ML)广泛应用于医学影像的分析,为癫痫的临床研究带来新的解决方案,尤其是在诊断、病灶定位和手术评估方面,提供了比传统方法更可靠、更精准的信息。因此,为了探索ML和神经影像学在癫痫研究中的可靠性和精确性,本篇综述我们总结了近几年ML在癫痫诊疗中的应用。首先,我们概述了癫痫研究中使用的典型神经影像技术和ML。然后,重点综述ML在以下三个研究热点中的应用:1)早期癫痫诊断;2)颞叶癫痫致痫灶的定侧;3)致痫灶的准确定位。最后,我们讨论了目前研究中取得的成果和面临的挑战,以及未来可能的研究方向,期望为癫痫的临床研究提供理论基础。

Epilepsy is a common neurological disorder,and its recurrent seizures seriously affect patient’s body and life.With the development of artificial intelligence,machine learning(ML)has been widely applied to the analysis of medical images,bringing new solutions to the clinical study of epilepsy,especially in diagnosis,lesion localization and surgical evaluation,and providing more reliable and accurate information than traditional methods.Therefore,to explore the reliability and accuracy of ML and neuroimaging in epilepsy research,in this review we summarize the application of ML in epilepsy diagnosis and treatment in recent years.First,we provide an overview of typical neuroimaging techniques and ML used in epilepsy research.Then,we focus on the review of the application of ML in the following three research hotspots:1)early epilepsy diagnosis;2)side of seizure focus in temporal lobe epilepsy;and 3)accurate localization of seizure focus.Finally,we discuss the achievements and challenges in the current research,as well as the possible future research directions,which are expected to provide a theoretical basis for clinical research in epilepsy.