Classification and Diagnosis of Lung Cancer Based Using CNN with VGG-19

Abstract

Lung cancer is a major contributor to cancer-related mortality globally, and timely identification is essential for enhancing patient prognosis. Recently, deep learning methods, specifically Convolutional Neural Networks (CNN), have demonstrated encouraging outcomes in image-based medical diagnosis. The paper suggests utilising a CNN-based method to diagnose lung cancer from a healthcare image dataset, with a specific focus on histopathological image data. The proposed CNN approach utilises the natural hierarchical characteristics found in healthcare imagery to autonomously acquire distinctive features that indicate lung cancer. Transfer learning from extensive image datasets and improving models that have been trained are used to tackle the limitations of limited healthcare image datasets successfully. The CNN model utilising the VGG-19 architecture is developed and tested on a comprehensive dataset of lung cancer patients. Following extensive testing and evaluation, the model demonstrates high accuracy as well as precision and recall in diagnosing lung cancer using medical imaging. Interpretability techniques are utilised to get insights into the model's decision-making process, hence increasing its transparency and therapeutic relevance. The proposed CNN-based technology has the potential to help radiologists and clinicians discover and diagnose lung cancer earlier, leading to better patient care and treatment outcomes.

Keywords: Lung Cancer, Histopathological image, Convolutional Neural Networks, VGG-19, Deep Learning

DOI: 10.7176/CEIS/15-1-04

Publication date: March 31^st 2024