Deep Learning Models for Classification of Lung Cancer Types from Histopathology Images

Abstract

the most deadly and often diagnosed cancer is lung cancer, among other types of cancers all over the world. being diagnosed early can save the patient's life and improve their five-year survival rate. In this context, accurately recognizing the types of lung cancer from histopathology images is essential because it helps doctors decide which cancer types need further therapy. In this paper, to identify types of lung cancer from histopathology images, a deep learning framework based on binary and multi-classification approaches has been proposed. The framework utilizes the concept of transfer learning, and the Weighted average ensemble approach is used for the multi-classification model. The performance of the proposed model is examined using the LC25000 dataset, which is made freely available, and compared with the current approaches for classifying cancer types. It is observed that for the binary classification problem, InceptionV3, EfficicientNetB1, and multi-classification using the Weighted average ensemble approach have provided better results. The figures of merit achieved are recall, average accuracy, precision, and F1-score of 98.88%, 98.77%,98.77%, and 98.77%, respectively © 2025 IEEE.