A Data Mining Approach to Detect Tuberculosis Using Clustering and GA-NN Techniques

During the past years, this has turn out to be clear that the range of Human Immunodeficiency Virus infection as well as persons immigrate from zones of high rate have ensued in bigger amount of Tuberculosis events. TB can affect all types of organs in a living being body. In previous years, TB classification has been done using various algorithms like color segmentation, thresholding, histogram equalization. The main objective of this research work is to create a data mining way out that makes identification of TB as exact as possible. In our proposed framework we have used various techniques such as centroid selection based clustering algorithm would be used to enhance the clustering scheme, PCA for feature extraction, genetic algorithm for feature optimization and neural network for training and testing purpose. In the end, results are being evaluated after classification and testing on the basis of performance parameter such as accuracy, recall, precision, false acceptance ratio, and false rejection ratio.