Intracerebral hematoma (ICH) is a disease with high mortality and poor prognosis rate, accounting for approximately 10% of all cerebrovascular disease. Manual extraction of ICH regions lacks accuracy and speed, and a quantitative evaluation method is needed. In this study, we propose a method that divides the extraction of ICH regions into multiple stages and extracts the target using two-class classification based on convolutional neural network. The performance of the model is evaluated using 18 subjects with intraventricular hemorrhage, and it is shown that the proposed method is promising for the extraction of ICH regions in a region with high absorption rates.
