Detection Of Cell Division Sequence Based-on Hidden Markov Model

With the development of computer technology, especially the image processing technology, computer image processing is widely uesed in the medical field. Through analysis the abnormal cells, morphological features extracted, the search of diagnosing the disease mechanism with the computer has attracted widespread attention. These studies can analyses medical images automatically, reduce the subjective interference, improve work efficiency, and reduce the burden on doctorsThe identifying of cell is an importance within medical science examination part, a lot of paroxysm is main dependent on medical science expert observe specimen inside the cell's appearance, proceeds to identify to cell with classification. This kind of artificial classification of work repetition but the monotone, efficiency is lowly. The article research the algorithm of detecting the abnormal cells based on Hidden Markov Model, can accurately and quickly identify the abnormal cells, and it is important.The thesis makes systemic research on three steps included in the algorithm. They are feature extraction, State division and detection of abnormal gene.For the feature extraction problem,for high dimensional data, the new feature selection method based on Fisher criterion and feature clustering is proposed. Firstly, the features that have more discrimination information based on Fisher criterion are selected. Then hierarchical clustering in the pre-selected subset is adopted. Finally the irrelevant and redundant features are removed. So that the efficiency of learning algorithm is enhanced and the computational complexity is reduced.According to the number of DNA and duration in every phase, the cell cycle is divided into different states.In the section of detection of abnormal gene, first, determines model parameters, and according to the model parameters using the Viterbi algorithm the feature data are divided into the most likely state sequence. Then amendments the initial model parameters, uses optimization formula of the parameters of Baum-Welch algorithm to optimize the parameters. Finally, the biggest expected algorithm of Baum-Welch algorithm is used to train the feature data. By this done, the Hidden Markov Model is obtained. Use this model to train. That is the method what we propose—detection of cell division sequence method based on support vector machine.