Advanced Computational Methods for Pelvic Bone Cancer Detection: Efficacy comparison of Convolutional Neural Networks
Jagbir Singh, P. R. Patel, Balaji Shesharao Ingole, Rambabu Inaganti, Vishnu Ramineni, Manjunatha Sughaturu Krishnappa, Bhushan Jayeshkumar Patel
Abstract
Pelvic bone cancer is a type of cancer that effects the hip(pelvis). Doctor use various methods to diagnose pelvic bone cancer, including imaging test like X-ray, CT scans, MRI scans, and PET scans. Sometimes, doctors take a small piece of bone called a biopsy to test and confirm the cancer type in a lab. However, the way doctors diagnose cancer is changing because of advancement of computers. Computers can help diagnose cancer more accurately and quickly. Therefore, this study contributes to novel methods for detecting bone cancer using computer algorithms by analyzing CT scan images from over 3,000 individuals with bone tumors. These images were categorized into two groups: one containing images with cancer and the other with images without cancer. We used Convolutional Neural Networks (CNNs)—to understand and analyze images. We tested five different CNNs to see which model is the best at finding bone sarcoma. Our study finds that the Xception model is the most effective at detecting bone cancer, with an accuracy of approximately 90.23%. This underscores the importance of selecting the appropriate neural network (NN) model for accurate cancer identification. This is a significant discovery in aiding doctors to detect cancer more quickly and accurately, potentially leading to earlier and more effective treatment for individuals with bone cancer.