Automated medical image classification using deep learning

Abstract

Medical imaging is extremely important in the domain of medicine. Image classification is now utilized to distinguish aberrant tissues from healthy tissue in brain imaging. The brain tumor is identified from MRI images by using some classification techniques, where the area of the tumor as well as the tumor size is detected. Automatic tumor detection using brain MRI is efficient and time- saving, assisting the neurologists in diagnosis. Tumors can increase the risk of cancer, which is the most common cause of death or major cause of mortality worldwide. To detect brain tumors at the moment, effective automation of tumor detection is essential. Marker based Watershed algorithm is a typical segmentation technique which is used for identifying brain tumors. For brain tumor detection, we performed marker based watershed classification on MRI images with the use of gray scale images, then by noise removal and morphological operations. The steps in the methodology are as follows: Gray-level and sharpening was used in the pre-processing, and the image was segmented using thresholding as well as the marker based watershed algorithm, and the CNN was used for classifying the images. Finally, the tumor's location and size were determined.