Abstract:
Lung cancer is the leading cause of cancer-related deaths worldwide. According to the
World Health Organization (WHO) 1.80 million people died in 2020 because of lung
cancer [1]. Lung cancer remains a leading cause of cancer-related mortality worldwide,
emphasizing the critical need for early and accurate detection methods. This research
paper introduces a novel approach to automated lung cancer cell detection utilizing an
advanced machine learning algorithm based on an analytical prediction method. We
utilized a Kaggle dataset named lung-and-colon-cancer-histopathological-images,
which encompasses three distinct classes, namely lung_aca (Lung Adenocarcinoma),
lung_n (Lung normal Tissue), lung_scc (Lung Squamous Cell Carcinoma) [2]. This
dataset was categorized based on these class attributes. Technology plays a pivotal role
in enhancing cancer detection methods, and numerous researchers have proposed
diverse approaches in this regard. In our study, we employed five classification models,
namely CNN, Xception, VGG16, ResNet-50, and Inception-v3, to identify early-stage
lung cancer (LC) using the provided dataset of histopathological image. The research
findings revealed that the VGG16 algorithm exhibited the highest classification
accuracy, achieving 99.35% for LC detection. In comparison, ResNet-50 achieved
99.26%, CNN attained 97.72%, Xception reached 89.79% and Inception-v3 achieved
85.25. These results underscore the significance of proper system design, tuning, and
the selection of machine learning methods in achieving accurate and efficient for
detection lung cancer in its early stages using clinical data.
