Research Article
Lung Cancer Diagnosis using a light weight deep learning model
@INPROCEEDINGS{10.4108/eai.5-1-2024.2341884, author={Mohit Agarwal and Vivek Mehta and Rohit Kr Kaliyar and Suneet Kumar Gupta}, title={Lung Cancer Diagnosis using a light weight deep learning model}, proceedings={Proceedings of the EAI 3rd International Conference on Intelligent Systems and Machine Learning, ICISML 2024, January 5-6, 2024, Pune, India}, publisher={EAI}, proceedings_a={ICISML}, year={2024}, month={8}, keywords={cnn compression lung cancer acceleration}, doi={10.4108/eai.5-1-2024.2341884} }- Mohit Agarwal
Vivek Mehta
Rohit Kr Kaliyar
Suneet Kumar Gupta
Year: 2024
Lung Cancer Diagnosis using a light weight deep learning model
ICISML
EAI
DOI: 10.4108/eai.5-1-2024.2341884
Abstract
Lung cancer is a disease in which lungs get infected by cancerous development of cells. This can be caused due to excessive smoking. However persons who do not smoke may also get the disease in today's polluted environment. The symptoms of lung cancer can be cough which does not cure, blood in cough, pain in chest, loosing weight. etc. The CT scans are used to diagnose type of cancer for their corresponding treatment. Generally lung cancer can be classified into 3 types of cancer: Adenocarcinoma, Squamous cell carcinoma, and Large cell carcinoma. To avoid any mis diagnosis machine learning and deep learning methods are very helpful to classify the exact type of cancer and whether it is present or not. Machine Learning methods such as Decision Trees (DT) and Random Forest (RF) gave very good performance with RF giving 97% accuracy. Similarly Convolution Neural Networks (CNN) such as Mobilenet and VGG19 were tested to give an accuracy of 78.12% and 81.25% respectively. A three layered CNN was also proposed to give an accuracy of 89%. Compressed MobileNet accuracy could be enhanced to 96.5%.
