Stroke is one of the common cerebrovascular diseases,which can be classified into acute(0-24 h),subacute(24 h-2 weeks)and chronic stage(>2 weeks).High-resolution T1-weighted imaging(T1WI)is one kind of magnetic resonance imaging(MRI)technique,which is main imaging mode of assessing the change of cerebral structure after stroke at chronic stage,and it can accurately judge the location and size of lesion.The precise segmentation for lesion is we accurately analyze dataset of neuroimaging in stroke with large scale,and is the key of predicting rehabilitation outcomes.The automatic segmentation algorithm is one kind of technique that can automatically segment data into many independent regions or object by specific regulation or model,which does not need manual intervention.Compared with manual segmentation,it has repeatable and scalable advantages,which does not need manual setting parameter.With the development of artificial intelligence technology,its application is gradually widespread in the field of medical imaging analysis.Adopting automatic segmentation algorithm to conduct reliable and repeatable segmentation for the lesions in T1WI,which can effectively assist clinical diagnosis and rehabilitation treatment for stroke.This review summarized the advance research of automatic segmentation algorithm based on T1WI for the lesion of chronic stroke from two aspects included conventional machine learning and deep learning,and the optimization approach of three kinds of algorithm structures,which can provide reference and inspiration for researching small lesion with higher difficulty in the field of automatic segmentation.

Stroke is one of the common cerebrovascular diseases,which can be classified into acute(0-24 h),subacute(24 h-2 weeks)and chronic stage(>2 weeks).High-resolution T1-weighted imaging(T1WI)is one kind of magnetic resonance imaging(MRI)technique,which is main imaging mode of assessing the change of cerebral structure after stroke at chronic stage,and it can accurately judge the location and size of lesion.The precise segmentation for lesion is we accurately analyze dataset of neuroimaging in stroke with large scale,and is the key of predicting rehabilitation outcomes.The automatic segmentation algorithm is one kind of technique that can automatically segment data into many independent regions or object by specific regulation or model,which does not need manual intervention.Compared with manual segmentation,it has repeatable and scalable advantages,which does not need manual setting parameter.With the development of artificial intelligence technology,its application is gradually widespread in the field of medical imaging analysis.Adopting automatic segmentation algorithm to conduct reliable and repeatable segmentation for the lesions in T1WI,which can effectively assist clinical diagnosis and rehabilitation treatment for stroke.This review summarized the advance research of automatic segmentation algorithm based on T1WI for the lesion of chronic stroke from two aspects included conventional machine learning and deep learning,and the optimization approach of three kinds of algorithm structures,which can provide reference and inspiration for researching small lesion with higher difficulty in the field of automatic segmentation.