Scholarlink

<p>The interruption of the blood supply to a particular part of the brain causes a brain stroke. This lack of blood flow can lead to the rapid death of brain cells, which results in brain damage, disability or even death. The early and acute detection of brain stroke lesions in Magnetic Resonance Imaging (MRI) images is crucial for effective treatment planning and patient outcomes. This study introduces a novel methodology for detecting brain strokes by leveraging a Siamese Convolutional Neural Network (SCNN) enhanced through optimization with the newly developed Fractional Black Winged Kite Algorithm (FBKA), aiming to improve diagnostic precision and computational efficiency. The methodology begins with acquisition of MRI images, followed by a denoising process using a Bilateral filter to enhance image quality. Stroke lesion segmentation is conducted using the Otsu thresholding technique, where the threshold values are optimally determined by FBKA, which combines the Black-winged Kite Algorithm (BKA) with a Fractional Concept. After segmentation, relevant features, such as Pyramid Histogram of Oriented Gradient (PHOG) and Local Vector Pattern (LVP), are extracted for further processing. The proposed SCNN-FBKA is employed for detecting brain stroke. The SCNN-FBKA shows better performance by achieving 95.765% of testing accuracy, Positive Predictive Value (PPV) of 92.868%, Negative Predictive Value (NPV) of 90.865%, False Positive Rate (FPR) of 0.056%, and False Negative Rate (FNR) of 0.066% for 90% of training data.</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

Optimized Thresholding and Siamese CNN for Brain Stroke Detection

  • 期刊论文

  • 发表日期：

    2026年3月14日

查看全文
  • Gopalsamy Venkadakrishnan Sriramakrishnan,
  • Ganeshan Ramasamy,
  • Padmavathi Pragada,
  • Sreenu Ponnada

摘要

The interruption of the blood supply to a particular part of the brain causes a brain stroke. This lack of blood flow can lead to the rapid death of brain cells, which results in brain damage, disability or even death. The early and acute detection of brain stroke lesions in Magnetic Resonance Imaging (MRI) images is crucial for effective treatment planning and patient outcomes. This study introduces a novel methodology for detecting brain strokes by leveraging a Siamese Convolutional Neural Network (SCNN) enhanced through optimization with the newly developed Fractional Black Winged Kite Algorithm (FBKA), aiming to improve diagnostic precision and computational efficiency. The methodology begins with acquisition of MRI images, followed by a denoising process using a Bilateral filter to enhance image quality. Stroke lesion segmentation is conducted using the Otsu thresholding technique, where the threshold values are optimally determined by FBKA, which combines the Black-winged Kite Algorithm (BKA) with a Fractional Concept. After segmentation, relevant features, such as Pyramid Histogram of Oriented Gradient (PHOG) and Local Vector Pattern (LVP), are extracted for further processing. The proposed SCNN-FBKA is employed for detecting brain stroke. The SCNN-FBKA shows better performance by achieving 95.765% of testing accuracy, Positive Predictive Value (PPV) of 92.868%, Negative Predictive Value (NPV) of 90.865%, False Positive Rate (FPR) of 0.056%, and False Negative Rate (FNR) of 0.066% for 90% of training data.