Scholarlink

<p>Early effective screening and grading of COVID-19 with Artificial Intelligence (AI) allows an accurate assessment of the extent and distribution of lung involvement. However, a precise demarcation of the lesions remains problematic due to their irregular structure and location(s) within the lung. We develop a novel AI-based framework for the automated segmentation of COVID-19 infected lung tissues from lung Computed Tomography (CT) scans. A novel encoder-decoder-based deep learning architecture, the Composite Deep Network with Feature Weighting <i>(CDNetFW)</i>, is proposed for efficient delineation of infected regions from lung CT images. Multiple conditional pathways are integrated into the encoder, thereby facilitating the discovery of robust and discriminatory characteristics. The novel feature weighting module helps prioritize the relevant feature maps to be probed along with those regions containing crucial information within these maps. This is followed by measuring the severity of the disease from the predicted segmentation maps. Comparative studies demonstrate the superiority of <i>CDNetFW</i> on publicly available lung CT datasets. It exhibits a mean Dice coefficient of 0.8145 and a mean Recall of 0.8216, indicating its ability to effectively detect true positive regions with a correspondingly reduced count of false negatives.</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

Composite Deep Network with Feature Weighting for Improved Delineation of COVID Infected Lung

  • 期刊论文

  • 发表日期：

    2026年7月7日

查看全文
  • Pallabi Dutta,
  • Sushmita Mitra

摘要

Early effective screening and grading of COVID-19 with Artificial Intelligence (AI) allows an accurate assessment of the extent and distribution of lung involvement. However, a precise demarcation of the lesions remains problematic due to their irregular structure and location(s) within the lung. We develop a novel AI-based framework for the automated segmentation of COVID-19 infected lung tissues from lung Computed Tomography (CT) scans. A novel encoder-decoder-based deep learning architecture, the Composite Deep Network with Feature Weighting (CDNetFW), is proposed for efficient delineation of infected regions from lung CT images. Multiple conditional pathways are integrated into the encoder, thereby facilitating the discovery of robust and discriminatory characteristics. The novel feature weighting module helps prioritize the relevant feature maps to be probed along with those regions containing crucial information within these maps. This is followed by measuring the severity of the disease from the predicted segmentation maps. Comparative studies demonstrate the superiority of CDNetFW on publicly available lung CT datasets. It exhibits a mean Dice coefficient of 0.8145 and a mean Recall of 0.8216, indicating its ability to effectively detect true positive regions with a correspondingly reduced count of false negatives.