Stroke is the most common and dangerous cerebrovascular disease. According to the statistics from World Health Organization (WHO), only following heart attack, stroke is one of the two leading causes of human deaths. In addition, in Vietnam, a shortage of specialized equipment and qualified professionals is becoming a significant problem for not only accurate diagnosis but also timely and effective treatment of stroke, especially intracranial hemorrhage (ICH), an acute case of stroke. This research will analyze challenges and show solutions for constructing an effective knowledge system in ICH diagnosis and treatment that helps to shorten professional gap among hospitals and regions. We suggest a service-oriented architecture for the big data driven knowledge system based on medical imaging of ICH. The architecture ensures the development of knowledge obeying a systematic and complete process including the exploration and exploitation of knowledge from medical imaging. Besides, the architecture adapts to modern trends in knowledge service modeling.