AI-Driven Lung Cancer Detection Using Deep Learning and Medical Imaging

Abstract

Lung malignancy remains one of the most poisonous diseases wide-reaching, and later detection significantly increases the chances of survival. Most of the time, the conventional diagnostic methods are based on the identification of lung cancer in advanced stages only when the treatment is less effective. This paper proposes a new hybrid deep learning technique that combines MobileNet, which is a lightweight Convolutional Neural Network (CNN), through Long Short-Term Memory (LSTM) networks on behalf of the accurate identification of lung cancer from medical tomography figures such as CT examinations and X-rays. MobileNet is used to automatically learn the most dis criminative features from the input images, and the LSTM layers focus on the sequential dependencies of these extracted features to further enhance the identification of very subtle patterns in early stage cancer. Besides, the proposed model is computationally efficient, thus suitable for mobile-based diagnostic applications in low-resource settings. The experimental results demonstrate the detection accuracy to be very high, thus the method can be a practical and effective real-time automated early-stage lung cancer diagnosis solution that would eventually lead to better clinical outcomes.