Lung cancer remains one of the most lethal types of cancer globally, emphasizing the necessity for early and accurate detection. This study suggests a deep learning-based classification model that uses the pretrained Xception model to distinguish between four different groups: normal lung tissue, squamous cell carcinoma, large cell carcinoma, and adenocarcinoma. We apply chest CT scan images of a chosen dataset meticulously to fine-tune the model utilizing image data improved by real-time preprocessing techniques. The model has a high accuracy rate, signifying its potential use in supporting clinical diagnosis. Grad-CAM (Gradient-weighted Class Activation Mapping) is used to map class-specific activation zones in order to improve explainability, which supports radiological diagnosis and makes explainable prediction poss ible. Our approach provides a promising AI-based solution to screen lung cancer with high accuracy and clinical interpretability.
Download PDF: https://toro.eu.org/RFqswI
