XSp-NET: XCEPTION SPINAL NETWORK FOR LEFT VENTRICULAR FAILURE CLASSIFICATION USING CARDIAC MRI IMAGE

Cardiac failure is a severe disease, which is generally found using the measurements of Left Ventricular Ejection Fraction (LVEF) using Magnetic Resonance Imaging (MRI) images. Even though MRI provides multiple images of the Left Ventricular (LV), which is effective in diagnosing cardiac failure, it consumes more time and cost. If the LV failure is not treated well, then it increases the mortality rate. In this research, Xception SpinalNet (XSp-Net) is developed to classify LV failure by combining the Xception module and SpinalNet. Initially, a cardiac MRI visual is given as input, and the Adaptive Weighted Median Filter denoises the MRI image in the pre-processing phase. The preprocessed visual is subjected to LV segmentation, which is done by the Fully Convolutional Network (FCN). Subsequently, the features, like Convolutional Neural Network (CNN), Complete Local Binary Pattern (CLBP), Local Optimal Oriented Pattern (LOOP), and Gray Level Co-occurrence Matrix (GLCM) are extracted from the image, which is segmented. These mined features are fed to the proposed XSp-Net to classify the LV failure. Furthermore, the highest accuracy, True Positive Rate (TPR), and True Negative Rate (TNR) recorded by XSp-Net are 91.04%, 92.10%, and 91.34% respectively for the learning set 90%.