In the application of Internet of things (IoT), the cerebrovascular disease detection e-health platform can realize the long care and long telemedicine of cerebrovascular health. The four levels architecture of the platform is designed and realized in the paper. The segmentation and reconstruction of brain vessel is deeply researched here. The double Gauss mixture model is put forward to realize the cerebrovascular segmentation and the stochastic estimation maximization (SEM) algorithm is adopted to estimate the parameters of it. It could be easily used for the common user of the platform without initial contour, high dimensional evolution equation and the iteration terminal condition. The cerebrovascular vessels have low proportion (<5%) in the brain tissue. Its angiography gray is non-uniform. Geometry is complex and individual differences are quite large. The segmentation method in our paper can get the good result. Ball B-Spline curve (BBSC) has the characters of strict mathematic foundation, less dataset, stable smoothness and continuity, good interactivity. It suits to transfer the data in the IoT platform. Combing these technologies, a CUDA (compute unified device architecture) based ray-casting volume rendering, the interactive and automatic virtual wandering are realized. The platform can realize the diagnosis of the cerebrovascular disease, the medical plan design and monitor in the treatment, which can also be widely used in the teaching. The research is an interesting try to extend the fine tissues e-health platform based on the internet of things.