Classification of Bone Tumor using Machine Learning and Deep Learning Approach

In this work, bone tumor is identified and classified in Magnetic Resonance (MR) images using mathematical models. MR images are subjected to preprocessing stage in which the original images are converted to grayscale image and morphological filters are used to enhance the tumor regions. The MR images are subjected to classification using support vector machines (SVM), convolutional neural network (CNN) and neural network classifiers (NN). Results show that the morphological filter has enhanced the tumor region from its surrounding tissues. SVM classifier has high accuracy compared to CNN and neural network classifiers. This may be due to the involvement of bone regions which has similar characteristics of lesions. Further the tumor is segmented and analyzed and it is observed that the convolutional neural network has increased accuracy compared to SVM classifier with minimal loss. Further this work can be extended by increasing the size of input for the analysis.