Optimized Hyperparameters Based Deep Neural Network for Lung Cancer Detection

Lung cancer is the cause of a higher percentage of death worldwide in recent times. Early accurate detection is essential for increasing survival rates, better treatment and public health. Very often, conventional diagnostic methods, including X-rays, CT scans, and biopsies, are time-consuming and need expert interpretation to detect accurately and efficiently. In recent times, medical diagnostics through imaging analysis have shown tremendous potential based on deep learning approaches, especially with the aid of Deep Neural Networks (DNNs). This study presents an advanced detection technique for lung cancer using an Optimized Deep Neural Network (DNN). The proposed methodology involves the generation of a synthetic dataset comprising various risk factors such as age, smoking history, genetic predisposition, air pollution exposure, and other relevant risk factors. Subsequently, Bayesian optimization is employed to achieve the optimal hyperparameters of the model for better performance. This results in a classification accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 8 \%}$</tex>, precision of 97.85 %, sensitivity of 98.40 %, and F1-Score of 98.12 %, showcasing this model as an innovative and advanced technique for automated lung cancer detection with higher accuracy, reliability and robustness. This research also includes explainability methods (SHAP, Grad-CAM), which will help the researchers to better understand how the model makes its decisions.