Hybrid Deep Learning and Machine Learning approach for Multi-Class Lung Cancer Diagnosis on CT Scans images

Abstract

Lung cancer remains a leading cause of mortality worldwide, necessitating early and accurate diagnostic approaches. This study presents a hybrid deep learning (DL) and machine learning (ML) framework for classifying lung cancer from CT scan images. The proposed methodology employs histogram equalization for contrast enhancement, followed by feature extraction using a pre-trained ResNet50 model, generating 1280-dimensional feature vectors. These features are subsequently classified using five machine learning algorithms: Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), XGBoost, and K-Nearest Neighbors (KNN). Experimental evaluation on a four-class CT dataset (adenocarcinoma, large cell carcinoma, squamous cell carcinoma, and normal tissue) demonstrates that Logistic Regression achieves superior performance with 90.00% accuracy, 90.35% precision, 90.00% recall, and 90.03% F1-score