Scholarlink

<p>Diagnostic systems constantly improve medical image analysis and diagnosis techniques to boost the speed and precision of cancer detection. One of the most difficult issues in classifying cervical cancer using deep learning algorithms is the accessibility and quality of data and changes in the images' size, shape, and physical appearance. This study presents important approaches for overcoming these hurdles and developing an improved diagnostic system. In this work, we utilize data from the Herlev dataset, including cervical cell images, disease symptoms, and their impact on cervical cancer classification and diagnosis. The data is pre-processed by employing imputation to handle missing values. Cropping, resizing, flipping, and rotation are common techniques to prepare images for analysis. Feature extraction using Stacked Bi-LSTM plays a vital role in deep learning by removing irrelevant and redundant features. A Spiking Neural Network with the Deer Hunting Optimization Algorithm (SNN-DHOA) is utilized for cervical cancer classification. The performance of the suggested technique is analysed using various metrics, achieving an accuracy of 98.21%, sensitivity of 98.27%, specificity of 98.15%, and an F1-score of 98.54%.</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

Automated cervical cancer classification using stacked BI-LSTM and optimized spiking neural network with deer hunting optimization algorithm

  • 期刊论文

  • 发表日期：

    2025年5月24日

查看全文
  • Harika Vanam,
  • G. Vijaylaxmi,
  • Vanam Sravan Kumar

摘要

Diagnostic systems constantly improve medical image analysis and diagnosis techniques to boost the speed and precision of cancer detection. One of the most difficult issues in classifying cervical cancer using deep learning algorithms is the accessibility and quality of data and changes in the images' size, shape, and physical appearance. This study presents important approaches for overcoming these hurdles and developing an improved diagnostic system. In this work, we utilize data from the Herlev dataset, including cervical cell images, disease symptoms, and their impact on cervical cancer classification and diagnosis. The data is pre-processed by employing imputation to handle missing values. Cropping, resizing, flipping, and rotation are common techniques to prepare images for analysis. Feature extraction using Stacked Bi-LSTM plays a vital role in deep learning by removing irrelevant and redundant features. A Spiking Neural Network with the Deer Hunting Optimization Algorithm (SNN-DHOA) is utilized for cervical cancer classification. The performance of the suggested technique is analysed using various metrics, achieving an accuracy of 98.21%, sensitivity of 98.27%, specificity of 98.15%, and an F1-score of 98.54%.