OVERALL SURVIVAL PREDICTION OF NSCLC USING RADIOMICS AND MACHINE LEARNING METHODS

Year 2024, Volume: 3 Issue: 1, 420 - 427, 29.07.2024

Muhammed Selman Erel , Hilal Arslan , Esra Şengün Ermeydan , İlyas Çankaya

Abstract

Cancer is one of the most fatal diseases. Millions of people all around the world die due to this illness as a result of abnormal cell growth. Billions of dollars are spent to cure and analyze it. Non-small cell lung cancer (NSCLC) is the most diagnosed type of lung cancer, which is a trending type of cancer. Accurate prognostic strategies are important for treating cancer patients. By this aim, radiomics is used to diagnose and prognose the disease in a non-invasive, budget-friendly, smart and fast way. In this study, 2-year survival prediction of NSCLC is performed by using radiomics and machine learning methods. Lung CT-scan images belonging to 422 patients retrieved from TCIA public DICOM archive are processed to detect meaningful features using open-source radiomics feature extractor, PyRadiomics. For classification step, K-Nearest Neighbor (KNN) and Support Vector Machines (SVM), Random Forest (RF) and Neural Network (NN) classifier methods are utilized with 10-fold cross validation. To achieve the best performance, the hyperparameters of machine learning methods are tuned using grid search method. Experimental results present that the NN achieves the best performance with an AUC score of 0.87, an accuracy of 0.81, a recall of 0.79 and an F-measure of 0.76.

Keywords

Cancer, NSCLC, Radiomics, KNN, SVM

References

• R. L. Siegel et. al., “Cancer statistics, 2018,” CA: A Cancer Journal for Clinicians, vol. 68, no. 1. Wiley, pp. 7–30, Jan. 2018. https://doi.org/10.3322/caac.21442
• S. Baek et al., “Deep segmentation networks predict survival of non-small cell lung cancer,” Scientific Reports, vol. 9, no. 1. Springer Science and Business Media LLC, Nov. 21, 2019. https://doi.org/10.1038/s41598-019-53461-2
• Parmar, C., Grossmann, P., Bussink, J., Lambin, P., & Aerts, H. J. W. L. (2015). Machine Learning methods for Quantitative Radiomic Biomarkers. In Scientific Reports (Vol. 5, Issue 1). Springer Science and Business Media LLC. https://doi.org/10.1038/srep13087
• Parmar, C., Leijenaar, R. T. H., Grossmann, P., Rios Velazquez, E., Bussink, J., Rietveld, D., Rietbergen, M. M., Haibe-Kains, B., Lambin, P., & Aerts, H. J. W. L. (2015). Radiomic feature clusters and Prognostic Signatures specific for Lung and Head & Neck cancer. In Scientific Reports (Vol. 5, Issue 1). Springer Science and Business Media LLC. https://doi.org/10.1038/srep11044
• Wu, W., Parmar, C., Grossmann, P., Quackenbush, J., Lambin, P., Bussink, J., Mak, R., & Aerts, H. J. W. L. (2016). Exploratory Study to Identify Radiomics Classifiers for Lung Cancer Histology. In Frontiers in Oncology (Vol. 6). Frontiers Media SA. https://doi.org/10.3389/fonc.2016.00071