Analysis and Design for the Early Stage Detection of Lung Diseases Using Machine Learning Algorithms

Machine learning (ML) is a fundamental method by creating complex, automated, and optimal methods by analyzing high-dimensional and multi-modal biomedical data. In medical systems, ML has importance to solve many health issues. This helps to diagnose infections faster and more effectively if they are identified earlier. This will save many lives and reduce the strain on the system. Computed Tomography (CT) photographs of the lungs include a variety of structures that are essential for diagnosing and analyzing pulmonary disease. Lung diseases like tuberculosis, pneumonia, repository diseases and lung cancer are all serious health risks. At present identification of lung cancers, lung diseases, pulmonary diseases, asthma, tuberculosis problems are major health issues and increasing rapidly. Many health issues are emerged in human body, lung diseases are most common and complex diseases leading to death in early stages. Lung diseases commonly occur due to loss of ability to breathe and lead to death due to late diagnosis. The lung diseases that occur in patients tend to play an important role but becoming difficult to detect the diseases in early stages in saving lives of patient in time. Physicians try to identify lung diseases by taking help of chest radiograph medical images which are one of the preliminary requirements for undergoing better diagnosis. Early detection of lung diseases can be helpful to patients for a better increasing in their health to cure and fast recovery. Due to rapid growth in the technology, the medial field has encounter dramatical changes and advancements in the way the diseases are simulated and analyzed efficiently. Researchers proposed various methods and approaches, but there is scope to improve the early detection of lung diseases via more efficient customized algorithms. In this approach, the objective is to improve the evaluation metrics and increase the diagnostic accuracy for early detection of lung diseases in patients. In this process, image segmentation methods are analyzed by applying various supervised and unsupervised machine learning algorithms and come up with better classification technique to design feature subsets stepwise. The features of lung medical images are extracted using statistical and geometrical methods for the detection of lung boundaries for better comparison of the results. Based on evaluation metrics like Mean Squared Error (MSE), Matthews Correlation Coefficient (MCC), Accuracy (A), Precision (P), Recall (R), Elapsed Time (ET), a comparison study is carried out by analyzing different ML techniques and finally results are demonstrated for the detection of lung diseases in early stages.