Increasing the efficiency of the Lesion segmentation tools to detect brain lesions in stroke

Stroke is the principal cause of death all over the world. Lesion detection is a crucial step in diagnosis and therapy of patients with stroke lesion. MRI is useful device in lesion segmentation that produces different brain image series, such as T1, T2 and FLAIR. Nowadays, automatic methods for detecting lesion are preferable to semi-automatic methods due to their speed, flexibility and availability. So, the efficiency of these methods in detecting lesion is important. LST-Lesion segmentation tool by SPM is an automatic tool which is most commonly used to detect Multiple Sclerosis (MS) lesions. Moreover, the LST has been investigated for detecting of stroke lesions. In this study the operation of this tool is examined both in the old version (SPM8) and the new one (SPM12) in a wide range of lesion volumes of ischemic stroke. The results show that LST can be used as an automated tool for detecting of stroke lesions but it works better in old version. Also, the performance of the LST in detection of lesions is improved by two feature maps. These feature maps separate missed and abnormal pixels. These two properties have more effects in detection of large lesions which are detected with high quality.