Scholarlink

<p>Lung Cancer (LC) is one of the most significant life-threatening cancers globally. Early detection and treatment are essential for patient recovery. Clinical professionals use histopathological images of biopsied lung tissue for diagnosis. However, identifying LC types is often error-prone and time-consuming. To address the above-mentioned challenges, an effective model Pyramid-KNet is developed for detecting LC using Computed Tomography (CT) images. Initially, the CT images are subjected to image enhancement using Histogram equalization. After that, lung lobe segmentation is completed by employing Psi-Net. Then, the lung nodule identification is conducted by the grid-based strategy. Furthermore, feature extraction is conducted for extracting features like Weber local descriptor (WLD), Median Robust Extended Local Binary Pattern with Discrete Cosine Transform (MRELBP with DCT), Texton, Gray-Level Co-occurrence Matrix (GLCM), and statistical features. Finally, LC detection is executed by employing hybrid Pyramid KroneckerNet (Pyramid-KNet), which is the integration of PyramidNet and Deep Kronecker Network (DKN), where layers are modified employing the Taylor concept. Furthermore, the performance of Pyramid-KNet is validated by comparing it with the performance of baseline methods, and Pyramid-KNet attained superior performance with an accuracy of 93%, precision of 92% and F-measure of 94% respectively.</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

Pyramid-KNet: Pyramid Kroneckernet-based lung cancer detection using computed tomography images

  • 期刊论文

  • 发表日期：

    2025年3月25日

查看全文
  • Praveen Kantha,
  • J. Anitha,
  • Katakam Venkateswara Rao,
  • Balajee Maram,
  • Satish Muppidi,
  • Parul Datta

摘要

Lung Cancer (LC) is one of the most significant life-threatening cancers globally. Early detection and treatment are essential for patient recovery. Clinical professionals use histopathological images of biopsied lung tissue for diagnosis. However, identifying LC types is often error-prone and time-consuming. To address the above-mentioned challenges, an effective model Pyramid-KNet is developed for detecting LC using Computed Tomography (CT) images. Initially, the CT images are subjected to image enhancement using Histogram equalization. After that, lung lobe segmentation is completed by employing Psi-Net. Then, the lung nodule identification is conducted by the grid-based strategy. Furthermore, feature extraction is conducted for extracting features like Weber local descriptor (WLD), Median Robust Extended Local Binary Pattern with Discrete Cosine Transform (MRELBP with DCT), Texton, Gray-Level Co-occurrence Matrix (GLCM), and statistical features. Finally, LC detection is executed by employing hybrid Pyramid KroneckerNet (Pyramid-KNet), which is the integration of PyramidNet and Deep Kronecker Network (DKN), where layers are modified employing the Taylor concept. Furthermore, the performance of Pyramid-KNet is validated by comparing it with the performance of baseline methods, and Pyramid-KNet attained superior performance with an accuracy of 93%, precision of 92% and F-measure of 94% respectively.