Scholarlink

One of the primary factors for death on a global scale is pneumonia as per World Health Organization. Conventional techniques for analyzing Chest X-ray (CXR) to diagnose pneumonia take a long time and call for specific medical knowledge, which increases the risk of delayed diagnosis and treatment. Immediate diagnosis and treatment of pneumonia can reduce the rate of mortality. Since CXR is a widely used, affordable, and accessible method of detecting pneumonia; a machine learning algorithm is presented in this paper to detect pneumonia using CXR. The proposed methodology extracts textural pattern features employing Completed Local Binary Pattern (CLBP) on CXR images and used those features as input to train the Convolutional Neural Network (CNN) models for classification. Experiment is carried out using a publicly available dataset named as Labeled Optical Coherence Tomography (OCT). CLBP images of CXR with proposed CNN model achieves the best average accuracy of 98.10%, when compared with other existing algorithms. In order to comprehend the classifier's decision-making process, the Gradient-weighted Class Activation Map is employed to map the essential features of CLBP images that are needed to predict the class using the proposed CNN model.

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

Interpretable Machine Learning Model to Detect Pneumonia from Completed Local Binary Pattern of Chest X-ray Images

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sushmita Pramanik Dutta,
  • Sriparna Saha,
  • Deepak Kumar,
  • Priyanka Pramod Pawar

摘要

One of the primary factors for death on a global scale is pneumonia as per World Health Organization. Conventional techniques for analyzing Chest X-ray (CXR) to diagnose pneumonia take a long time and call for specific medical knowledge, which increases the risk of delayed diagnosis and treatment. Immediate diagnosis and treatment of pneumonia can reduce the rate of mortality. Since CXR is a widely used, affordable, and accessible method of detecting pneumonia; a machine learning algorithm is presented in this paper to detect pneumonia using CXR. The proposed methodology extracts textural pattern features employing Completed Local Binary Pattern (CLBP) on CXR images and used those features as input to train the Convolutional Neural Network (CNN) models for classification. Experiment is carried out using a publicly available dataset named as Labeled Optical Coherence Tomography (OCT). CLBP images of CXR with proposed CNN model achieves the best average accuracy of 98.10%, when compared with other existing algorithms. In order to comprehend the classifier's decision-making process, the Gradient-weighted Class Activation Map is employed to map the essential features of CLBP images that are needed to predict the class using the proposed CNN model.