Enhancing Lung Cancer Classification with MobileNetV3 and EfficientNetB7: A Transfer Learning Approach

Lung cancer presents a major global health challenge, requiring swift and precise detection for effective treatment. This study examined the effectiveness of six transfer learning models-MobileNetV3, DenseNet201, EfficientNetB7, VGG16, VGG19, and Inception V3-using histopathological images. The dataset included 15,000 images of various lung cancer types, aiming to categorize them into adenocarcinoma, benign, or squamous cell carcinoma. Metrics such as accuracy, precision, recall, and F1-score were used for evaluation. MobileNetV3 achieved an exceptional accuracy of 99.70%, outperforming Inception V3 (78%), DenseNet201 (93%), VGG16 (99%), VGG19 (98%), and EfficientNetB7 (99.50%). Both MobileNetV3 and EfficientNetB7 excelled in accuracy, precision, recall, and F1- score. Therefore, this study recommends MobileNetV3 and Effi- cientNetB7 for lung cancer classification. The proposed method demonstrated significant accuracy, underscoring its effectiveness and practical use. Key model features showed robust precision, recall, and F1-score metrics at 100%, 99%, and 100%, respectively the highest among the models compared. The study also developed a Streamlit-based web application for users to upload images for lung cancer prediction. This research adds to the existing literature by offering a comprehensive approach to early- stage lung cancer detection, addressing previous study limitations and current knowledge gaps.