Scholarlink

<p>The process of retrieving identical medical images in large databases is the Content-Based Medical Image Retrieval (CBMIR). Nowadays, the requirement for image storage is increasing to petabytes. In such a huge volume of data, the existing methods consume more time to search and retrieve identical images with less accuracy. To address these troubles, this paper designs a novel Enhanced residual Network with dilated Content-driven attention and an Oppositional Levy flight-based Grasshopper optimization algorithm (ERDCA-OLFGOA) that focus on retrieving identical images depending on the query images. Initially, the input image is pre-processed to improve the image’s quality. The ERDCA-OLFGOA model applies discrete wavelet transformation for extracting the input image features. The enhanced ResNet50 with dilated content-driven attention captures the hierarchical feature and multi-scale contextual information in the medical images. To enhance the retrieval performance, ERDCA-OLFGOA model implements Oppositional Levy flight-based Grasshopper optimization algorithm for hyperparameter tuning of enhanced ResNet50 structure. Finally, the Euclidean distance examines the similarity that found between the query and database images and then retrieves the identical images. On validation, the proposed ERDCA-OLFGOA technique achieved better performance and provides more accurate and effective results in the CBMIR task.</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

Enhanced ResNet50 with DWT-based CBMIR strategy: ERDCA-OLFGOA framework

  • 期刊论文

  • 发表日期：

    2026年6月20日

查看全文
  • Sivasakthi Thanigainathan,
  • Palani Uthirapathy

摘要

The process of retrieving identical medical images in large databases is the Content-Based Medical Image Retrieval (CBMIR). Nowadays, the requirement for image storage is increasing to petabytes. In such a huge volume of data, the existing methods consume more time to search and retrieve identical images with less accuracy. To address these troubles, this paper designs a novel Enhanced residual Network with dilated Content-driven attention and an Oppositional Levy flight-based Grasshopper optimization algorithm (ERDCA-OLFGOA) that focus on retrieving identical images depending on the query images. Initially, the input image is pre-processed to improve the image’s quality. The ERDCA-OLFGOA model applies discrete wavelet transformation for extracting the input image features. The enhanced ResNet50 with dilated content-driven attention captures the hierarchical feature and multi-scale contextual information in the medical images. To enhance the retrieval performance, ERDCA-OLFGOA model implements Oppositional Levy flight-based Grasshopper optimization algorithm for hyperparameter tuning of enhanced ResNet50 structure. Finally, the Euclidean distance examines the similarity that found between the query and database images and then retrieves the identical images. On validation, the proposed ERDCA-OLFGOA technique achieved better performance and provides more accurate and effective results in the CBMIR task.