Automated Severity Analysis of TB Using Center Based Genetic Neural Network

In the medical image analysis Classification plays a crucial role especially in disease diagnosis. It helps physicians and radiologists in their decision interpretations about the disease. Tuberculosis (TB) is an infectious disease arised by Mycobacterium tuberculosis. Lot of researchers accomplished work in TB diagnosis. Still TB detection is a key problem. Accordingly an accurate TB detection process is suggested using optimal neural network classifier. In our work Initially the feature vector will be drawn out from the given dataset by computing First order statistical, structural and gray level dependent features for all input images present in the database. Later the feature selection process will be done. On that chosen features can be executed to train the Neural Network classifier. In the proposed system, we use center based genetic algorithm (CBGA) in order to optimize the weights of Network. It classifies the given image as abnormal or normal. The proposed work is implemented using the MATLAB. Compared to existed methods, performance of our projected system is enhanced in respect of Sensitivity, Specificity and Accuracy.