Scholarlink

<p>In medical Chest Radiography (CXR) and Computed Tomography (CT) imaging, the critical importance of accurate data acquisition offers significant challenges in multi-class classification. This study provides an exhaustive examination of a variety of lung-related diseases, including the recent Covid-19 pandemic, and highlights their prevalent symptoms. Additionally, we emphasize significant contributions from the research community and identify the most popular methodologies for addressing the multi-class classification conundrum. We conduct exhaustive experiments to further our comprehension of the results of these experiments, drawing on these insights. Our methodology is based on a seven-class classification approach, which produces exceptional results. Specifically, our concatenation-based method on balanced data obtains an F1-score of 97.56% and an impressive accuracy of 97.57%. Our framework maintains high performance, obtaining an F1-score of 94.64% and 95.22% accuracy on unbalanced data. These results are meticulously compared to state-of-the-art predictive models, thereby enhancing the existing literature with empirical evidence. Additionally, we investigate the integration of a variety of modalities, including CT and CXR images, with the objective of enhancing the efficacy of the model. Our results emphasize the efficacy of the proposed predictive models in CXR imaging, offering practical insights for both researchers and practitioners.</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

Enhancing Predictive Modeling in Using CT and CXR Images: A Review of Lung Disease Detection and Experimental Insights

  • 期刊论文

  • 发表日期：

    2025年5月29日

查看全文
  • Manojeet Roy,
  • Ujwala Baruah,
  • Shashank Sharma,
  • Priyam Dey,
  • Mansi Rawat,
  • Barnika Paul

摘要

In medical Chest Radiography (CXR) and Computed Tomography (CT) imaging, the critical importance of accurate data acquisition offers significant challenges in multi-class classification. This study provides an exhaustive examination of a variety of lung-related diseases, including the recent Covid-19 pandemic, and highlights their prevalent symptoms. Additionally, we emphasize significant contributions from the research community and identify the most popular methodologies for addressing the multi-class classification conundrum. We conduct exhaustive experiments to further our comprehension of the results of these experiments, drawing on these insights. Our methodology is based on a seven-class classification approach, which produces exceptional results. Specifically, our concatenation-based method on balanced data obtains an F1-score of 97.56% and an impressive accuracy of 97.57%. Our framework maintains high performance, obtaining an F1-score of 94.64% and 95.22% accuracy on unbalanced data. These results are meticulously compared to state-of-the-art predictive models, thereby enhancing the existing literature with empirical evidence. Additionally, we investigate the integration of a variety of modalities, including CT and CXR images, with the objective of enhancing the efficacy of the model. Our results emphasize the efficacy of the proposed predictive models in CXR imaging, offering practical insights for both researchers and practitioners.