Publication

Abstract : Bone cancer is an epidemic of the skeletal system that is impacted by a complex interaction of genetic variables and inherited susceptibility. It is critical to develop a readily accessible detection and classification method for early detection of this cancer. As a consequence, a computerized procedure for distinguishing between cancerous and healthy bones is necessary. This paper is dedicated to developing a system that uses multiple magnetic resonance imaging (MRI) scans in digital imaging and communications (DICOM) format of cancerous and healthy bones that exhibit comparable structural traits from different individuals to identify and classify bone cancer using convolutional neural networks (CNN). The proposed approach can be broken down into three parts: MR image improvement, feature encoding, and categorizing. After preprocessing an image, the features are retrieved using discrete wavelet transformation (DWT). Finally, neural networks (NN) differentiate between normal and diseased bone. This method hybridizes one of the most efficient CNN architectures, Visual Geometry Group (VGG) 19. An effective strategy for cancer detection is proposed based on spatial fuzzy C-means (SFCM) clustering. A rigorous performance evaluation showed that the proposed CBCD method is extremely successful in detecting bone cancer, with an amazing accuracy rate of 92.9%.