| （英） |
Brain tumor segmentation is one of the systems that a computer, which has attracted attention in recent years, assists doctors in diagnosis. Conventionally, the mainstream method is to obtain a threshold value for judging whether a tumor is a tumor from the pixel values of the brain image.In this study, we proposed a two-level hidden Markov model to express the mechanism of brain tumor development and the physical structure of the human brain, and expressed brain MRI with tumor. In this way, it is possible to make use of the data to make use of the background knowledge of the data to solve various problems.
In this study, this model was formulated as a state estimation problem, and the optimal decision was derived under Bayesian criteria. After that, we derived an efficient approximation algorithm. |