Lung Cancer Types Prediction Using Machine Learning Approach

• Main Authors: Ingle, Kshitija, Chaskar, Uttam, Rathod, Shashikant
• Format: Conference Proceeding
• Language: English
• Subjects: AdaBoost; Computed tomography; Feature extraction; Lung; Lung cancer; Machine Learning algorithm; Machine learning algorithms; Sensitivity; Support vector machines
• ISSN: 2766-2101
• DOI: 10.1109/CONECCT52877.2021.9622568

Lung cancer is the most common type of cancer in India, along with prostate, mouth, breast cancer. Due to smoking, excessive increases in pollution and the inhalation of cancerous elements cause cancer in men and women. Still, the amount of cancer in men is more compared to women. Different imaging modalities are used for diagnoses, such as CT, MRI, PET, and X-Ray. Oncologists and physicians diagnose cancer by seeing gray-scale saturation in the image. It requires experience in both anatomy and physiology of humans with imaging technique knowledge. It takes too much time for diagnosing, and accuracy is less due to a lack of imaging expertise. To avoid time consumption and provide easy diagnosis, the researcher has used an AI-based application technique for diagnosis purposes. In this study, we have used the Ensemble Machine Learning algorithm, i.e., AdaBoost, to predict different lung cancer types. We trained ensemble learners using features extracted from the lung CT images and evaluations done using performance metrics. The Accuracy of Adaboost is 90.74%, with sensitivity is 81.80%, specificity is 93.99%, F1 score 0.8, kappa is 0.753, and AUC is 0.93. The performance of the AdaBoost classifier is compared with the different machine learning algorithms.