Classification of Brain MRI Volume Data Based on Wavelet-Domain Hidden Markov Model
Xueyan Lin · Journal of Tianjin University Science and Technology · 2007
To ensure the 3D visualization to render the tissues accurately, a new classification of brain magnet-ism resonance imaging volume data was proposed based on wavelet-domain hidden Markov model. First, param-eters of HMT were estimated by expectation-maximization (EM) algorithm. Then the approximative class and the eigenvalue of each class in wavelet space, which could be obtained only by lots of training using the existent algorithms, were gained by proper wavelet-decomposition. The result of classification was obtained by iterated conditional mode algorithm, which shows that the proposed algorithm is more accurate but less complex.