Bone Cancer Cell Prediction Using an Enhanced Deep Learning Algorithm with an Optimization Technique.

Area of Science:

• Medical imaging analysis
• Computational oncology
• Machine learning in healthcare

Background:

• Early detection of bone cancer is critical for patient survival.
• Traditional diagnostic methods are time-consuming and require specialized expertise.
• Automated systems are needed to improve the accuracy and efficiency of bone cancer diagnosis.

Purpose of the Study:

• To develop a machine learning (ML)-driven tool for enhanced bone cancer detection and classification.
• To improve diagnostic accuracy by integrating deep learning (DL) with optimization algorithms.
• To create a clinically applicable model for early bone cancer diagnosis.

Main Methods:

• Utilized a hybrid approach combining ResNet with Cuckoo Search Modified Hill Climbing (CS-MHC) optimization.
• Employed Cuckoo Search Optimization (CSO) for feature selection and hyperparameter tuning.
• Compared the CS-MHC ResNet model against traditional DL models (VGG-16, Inception, Xception).

Main Results:

• CS-MHC ResNet achieved superior classification accuracy (over 90%), sensitivity (around 85%), precision (over 88%), and F-measure (approx. 86%).
• The integrated CSO enhanced feature selection and improved bone cancer classification effectiveness.
• Outperformed traditional models, demonstrating the efficacy of the hybrid optimization and DL approach.

Conclusions:

• The CS-MHC ResNet model offers a significant advancement in automated bone cancer detection.
• The model demonstrates high potential for clinical application, providing a more efficient and reliable diagnostic tool.
• Future work will focus on validation with larger datasets and exploring simpler model designs for broader applicability.