This computer-aided diagnosis (CAD) software (MATLAB toolbox) has been developed for automated prediction of tuberculosis (TB) from chest X-ray (CXRs) of patients. This toolbox was developed by incorporating more diversified global features extraction methods such as Gist and PHOG. It is effective in discriminating between CXR(s) of non-TB and TB patients. It contains two modules: Training and Prediction Modules. The latter (Prediction Module) predicts input digital CXR(s) image as TB or non-TB so it will useful for general user. The former module (Training module) enables user to develop a model trained on his/her own TB and nonTB chest radiographs. The user developed automated model could be used for detection of TB from CXR(s). User has also option to choose between Gist and PHOG features to be used in both training and prediction module. This TB-Xpredict software is extremely efficient and having a maximum prediction accuracy of 94.2% with sensitivity of 0.961 for Gist features.
