Abstract
Bone cancer is a malignant tumor that affects the healthiest tissues in the bone. Bone cancer is identified by swelling, bone weakness risk factors, lumps in the affected area, fever, chills, and night sweat symptoms. Despite the fact that bone cancer produces significant symptoms, it is difficult to predict in beginning stages because of the low priority of its symptoms. Several optimization techniques, such as medical image analysis and machine learning techniques, have been utilized to detect the initial stages of bone cancer. These methods sometimes fail to accurately predict bone cancer because of the error rate and complexity of the tissue structure. In this work, we introduce particle swarm optimized extreme learning neural networks for effectively predicting bone cancer. Initially, X-ray images are gathered from the oral cancer database, that must be examined noise to eliminate with the assistance of a non-local median filter. Then, the cancer affected region is segmented with the help of an enhanced multi-scale segmentation algorithm, and features are extracted from the identified region. The extracted features are classified using Particle Swarm based Extreme Learning Neural Networks Classifier. The introduced technique is superior to the current known classifier and could 98.2% accuracy which is obtained from MATLAB based experimental results.