Automated Bone Cancer Detection Using Deep Learning on X-Ray Images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Bone cancer diagnosis relies on medical imaging but faces challenges like high costs, time consumption, and potential misdiagnosis.
• Automated systems are crucial for precise and efficient bone cancer detection to aid physicians.
• Deep learning shows promise in enhancing medical image analysis for cancer classification.

Purpose of the Study:

• To develop and evaluate an automated system for classifying bone X-ray images as healthy or cancerous.
• To improve the accuracy and efficiency of bone cancer diagnosis using artificial intelligence.
• To reduce human labor and minimize diagnostic errors in bone cancer screening.

Main Methods:

• A novel Golden Search Optimization along with Deep Learning Enabled Computer Aided Diagnosis for Bone Cancer Classification (GSODL-CADBCC) approach was developed.
• The technique utilizes bilateral filtering for noise reduction, SqueezeNet for feature extraction, and Golden Search Optimization (GSO) for hyperparameter selection.
• Classification is performed using an improved cuckoo search with a long short-term memory (LSTM) model.

Main Results:

• The GSODL-CADBCC approach achieved an average accuracy of 95.52% on the training dataset.
• The system demonstrated a testing accuracy of 94.79% on unseen X-ray images.
• The automated method significantly reduced manual interpretation needs and diagnostic error risks.

Conclusions:

• The developed GSODL-CADBCC system offers a precise and automated solution for bone cancer screening using X-ray images.
• This AI-driven approach enhances diagnostic accuracy and efficiency in identifying cancerous bones.
• The study highlights the potential of deep learning and optimization algorithms in medical image analysis for life-threatening diseases.