Detection of Lung Cancer Malignancy Types on CT-Scan Using the Convolutional Neural Network Method at PHC Hospital Surabaya

Abstract

There are many uses for digital image processing, ranging from tumor and cancer detection in the body to reading blood cells. The rate of lung cancer represents about 13.27% of the total cancer cases, and this shows that lung cancer is the main type of disease in men. Lung cancer is one of the most dangerous and life-threatening diseases in the world. In Indonesia, lung cancer is more often detected when patients are at an advanced stage. Therefore, in this paper, we applied Deep Learning to solve a lung cancer malignant detection system; it is used to detect and classify nodule areas. So that lung cancer detection can be obtained with accurate results. This paper explains the working system for detecting lung cancer malignancies using a Convolutional Neural Network (CNN) and the model architecture for training the dataset using the EfficientNet model. This study collected 800 lung CT images from PHC Surabaya Hospital in DICOM format. A total of 13 layers with EfficientNet architecture and classification layers for each type of cancer class have been used in the model. The experimental results of the model achieved satisfactory results with an accuracy of 99.46%, with a maximum epoch of 30 and a mini-batch size of 128.