LuNet: Development of Lung Cancer detection from CT Scan images using hybrid Customized deep Convolution Neural Networks

• Pradeepa Sampath,
• Sasikaladevi Natarajan,
• S. Vimal,
• Sanghyun Seo,

Vol. 19, No. 10, October 31, 2025
10.3837/tiis.2025.10.002, Download Paper (Free):

• Lung cancer Image Classification
• Hybrid CNN
• ResNet 50
• Customized CNN
• CT Scan Images

Abstract

In the realm of human existence, various factors contribute to mortality in humans, with cancer being one of the most perilous conditions. The lung cancer is the main reason of the death in all over the world. A Computed Tomography (CT) scan is used to determine the tumour’s location and the extent of malignancy within the body. We proposed a novel hybrid customized deep convolution neural network (LuNet) is used to identify the cancerous cells in the lung with Computed Tomography scan images. Bench mark data is collected from the Kaggle and pre-processed using Min Max normalization techniques. The essential features are extracted from pre trained customized neural network such as customized ResNet50 and customized CNN. Both the features are concatenated through the fully connected layer and classified using Softmax Layer with three different classes such as Benign, Malignant, and Normal tumours. This integrated approach aims to identify small cancer-infected cells from CT scan images, potentially saving lives through targeted prevention, early detection, and treatment strategies. The proposed methodology achieves 98.91% accuracy, which is higher compare to other state of art algorithms.

Statistics

Cite this article