Scholarlink

<p>Emphysema and chronic bronchitis are among the most common lung conditions classified under Chronic Obstructive Pulmonary Disease (COPD). COPD is usually caused by habits like smoking, chewing tobacco, or long-term exposure to harmful irritants. By 2030, it is estimated that emphysema will significantly affect 10% of the global population. Early diagnosis of emphysema is essential to ensure timely treatment and intervention. This study introduces an innovative Machine Learning (ML)-based method: a stacking-based classifier for classifying emphysema disease via Histogram of Oriented Gradients (HOG) and Principal Component Analysis (PCA)-Enhanced Computed Tomography (CT) imaging. Three datasets, including one online and 2 local hospital datasets, were integrated, pre-processed, and balanced using Borderline SMOTE for improved class distribution. The HOG method was used for feature extraction, and dimensionality reduction was performed with PCA. Various ML models such as Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), and Naive Bayes (NB) were analyzed, with RF and SVM delivering better results. A stacking classifier was developed to enhance classification accuracy by combining RF and SVM as base models and logistic regression as the meta-model. The proposed model achieved a validation accuracy of 97.89% and a balanced testing accuracy of 97.90%, demonstrating its effectiveness in classifying emphysema. This approach addresses the challenges of emphysema disease classification and shows great potential for clinical applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Classification of emphysema disease using stacking classifier through computed tomography images

  • 期刊论文

  • 发表日期：

    2025年12月27日

查看全文
  • Ragipati Karthik,
  • R Murugan,
  • Tripti Goel

摘要

Emphysema and chronic bronchitis are among the most common lung conditions classified under Chronic Obstructive Pulmonary Disease (COPD). COPD is usually caused by habits like smoking, chewing tobacco, or long-term exposure to harmful irritants. By 2030, it is estimated that emphysema will significantly affect 10% of the global population. Early diagnosis of emphysema is essential to ensure timely treatment and intervention. This study introduces an innovative Machine Learning (ML)-based method: a stacking-based classifier for classifying emphysema disease via Histogram of Oriented Gradients (HOG) and Principal Component Analysis (PCA)-Enhanced Computed Tomography (CT) imaging. Three datasets, including one online and 2 local hospital datasets, were integrated, pre-processed, and balanced using Borderline SMOTE for improved class distribution. The HOG method was used for feature extraction, and dimensionality reduction was performed with PCA. Various ML models such as Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), and Naive Bayes (NB) were analyzed, with RF and SVM delivering better results. A stacking classifier was developed to enhance classification accuracy by combining RF and SVM as base models and logistic regression as the meta-model. The proposed model achieved a validation accuracy of 97.89% and a balanced testing accuracy of 97.90%, demonstrating its effectiveness in classifying emphysema. This approach addresses the challenges of emphysema disease classification and shows great potential for clinical applications.