A parallel processing model for big medical image data

Parallel computing has gained a great influence on scientific researches and in our daily life, especially when dealing with big data. One of the preconditions of high performance on computing is the support of efficient algorithms, which should be divisible and computing simultaneously. But not all algorithms are applicable for parallel computing, sometimes it can only make use of one single processor. In order to take full advantages of cluster or Multi-core CPUs in that case, A pipeline computation model is proposed which applies on cluster to make procedures more efficient and make full use of computer resources. Especially, our model has a very good performance on medical image process. With the model, almost all the positions of the organs in CT-images of a person could be found out simultaneously and accurately in one time, which can efficiently speed up the diagnosis of doctors, rather than the serial algorithm which can only find the position of one organ in one time before. The result of our experiment shows that the performance of the former serial algorithm has been improved by 40 percent by using our method.