Scholarlink

<p>Brain stroke is one of the leading causes of neurological disability and imaging-based early identification is essential for effective diagnosis. However, existing models fail to effectively balance noise suppression with the preservation of critical lesion boundaries. In this research work, a novel CARE-MRI approach has been proposed for denoising the brain MRI images to detect brain stroke in its early stages by enhancing the image quality. This model introduced a Dual Channel-Hint Residual Network integrated with two specific blocks, Channel Attention Residual (CAR) Block, and Hint Pixel-Based Residual (HPR) Block for effectively removing noise. The CAR Block improves feature extraction by prioritizing the most informative channels through a channel-wise attention mechanism. The HPR Block enhances pixel-level detail utilizing a computationally efficient four-corner pixel approach. The two-stage denoising process ensures progressive refinement of noisy images without compromising on structural integrity. These components ensure the preservation of structural details, such as lesion boundaries and tissue contrast, while reducing computational complexity. The effectiveness of the CARE-MRI approach was assessed using Peak Signal-to-Noise Ratio (PSNR), Mean Square Error (MSE), Feature Similarity Index Measure (FSIM), and Structure Similarity Index Measure (SSIM). The CARE-MRI approach attains a high PSNR value of 38.3 and SSIM value of 0.9927 for brain MRI image denoising. The proposed approach improves the total PSNR value of 7.1, 5.8, 5.2, and 3.8 compared to LAN, DenseTrans, F2Net, and DRANet, 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

CARE-MRI: Dual Channel-Hint Attention Block Integrated Residual Network for Brain MRI Image Denoising

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月17日

查看全文
  • J. Sworna Jo Lijha,
  • N. Muthukumaran

摘要

Brain stroke is one of the leading causes of neurological disability and imaging-based early identification is essential for effective diagnosis. However, existing models fail to effectively balance noise suppression with the preservation of critical lesion boundaries. In this research work, a novel CARE-MRI approach has been proposed for denoising the brain MRI images to detect brain stroke in its early stages by enhancing the image quality. This model introduced a Dual Channel-Hint Residual Network integrated with two specific blocks, Channel Attention Residual (CAR) Block, and Hint Pixel-Based Residual (HPR) Block for effectively removing noise. The CAR Block improves feature extraction by prioritizing the most informative channels through a channel-wise attention mechanism. The HPR Block enhances pixel-level detail utilizing a computationally efficient four-corner pixel approach. The two-stage denoising process ensures progressive refinement of noisy images without compromising on structural integrity. These components ensure the preservation of structural details, such as lesion boundaries and tissue contrast, while reducing computational complexity. The effectiveness of the CARE-MRI approach was assessed using Peak Signal-to-Noise Ratio (PSNR), Mean Square Error (MSE), Feature Similarity Index Measure (FSIM), and Structure Similarity Index Measure (SSIM). The CARE-MRI approach attains a high PSNR value of 38.3 and SSIM value of 0.9927 for brain MRI image denoising. The proposed approach improves the total PSNR value of 7.1, 5.8, 5.2, and 3.8 compared to LAN, DenseTrans, F2Net, and DRANet, respectively.