Bone Mineral Density Measurement for Detection of Bone Cancer Using Recurrent

Abstract

Recently, the number of people affected by bone cancer is increasing at a rapid rate. This type of cancer can be cured easily, if identified at earlier stages. Modern imaging techniques are popularly used for the diagnosis of bone cancer. In this research, we propose a new methodology for the identification of bone cancer. In this methodology, the first step is the acquisition of bone images using imaging techniques. The next step is the filtration using 2D Hybrid Bilateral filter. The third step is enhancement using proposed Edge Preservation based Contrast and Brightness Equalization (EPCBE) algorithm. This algorithm was designed such that histogram equalization was done with the aim to preserve the edge information of the images. The next step is the clustering which is performed using Improved Fuzzy C Expectation Maximization (IFCEM) algorithm and thresholding is done using adaptive Otsu (AO) thresholding algorithm. Then, the Grey level Co-occurrence matrix (GLCM) and Grey level difference method (GLDM) matrices are extracted. From these matrices, various statistical features are extracted. Finally, the classification is done using RNN network architecture. The proposed methodology achieved robust results in terms of various evaluation metrics like mean square error, Jaccard coefficient, specificity, recall etc. In particular, the proposed scheme achieved minimal mean square error of 0.87, high Jaccard coefficient of 0.761, high specificity of 96.25% and highest recall rate of 96.52%.