Glioma brain tumor detection using dual convolutional neural networks and histogram density segmentation algorithm

In this work, Glioma brain tumor images are detected from the healthy brain images using Edge preserving image fusion and dual-deep learning Convolutional Neural Network (CNN) method. The first CNN module is used to extract the internal features from the brain image and the second CNN module is used for the feature classification process in this work. In case of training, the brain tumor images and the healthy brain images from the BraTS-IXI dataset are fused using edge preserving method and the fused images are data augmented for increasing the image counts for classification process. Then, the data augmented images are classified by the proposed CNN classifier (to be functioned in training) to produce the Trained Vector (TV). In case of testing process, the source brain images are fused with edge preserving method and the fused image is data augmented and further these data augmented images are classified (to be functioned in classification) into either Glioma or healthy brain image. Then, Histogram-Density Segmentation Algorithm (HDSA) is proposed to segment the tumor regions in the classified Glioma images.