Scholarlink

Background <p>Artificial intelligence is being developed as a vital tool to improve the accuracy and efficiency of oral cancer diagnosis. Oral squamous cell carcinoma is often detected at advanced stages, leading to poor patient outcomes, and traditional diagnostic methods can be subjective and physically demanding for pathologists. By comparing different Artificial intelligence models, including machine learning and deep learning, this study aims to create a more standardized and rapid diagnostic aid that can support pathologists and increase patient survival rates.</p> Methods <p>A total of 115 cases (49 OSCC, 66 normal) were captured at 100x and 400x magnifications. Twelve Gray Level Co-occurrence Matrix (GLCM) features were extracted. Machine learning models, including Fine Tree, Medium Tree, Linear Discriminant, Linear Support Vector Machine, K-Nearest Neighbors, and Neural Network (MLP) model were developed. Model performance was evaluated across four distinct classification tasks, assessing the contribution of histopathological images alone and in fusion with demographic/clinical data. Statistical validation involved 5-fold cross-validation on the training set and independent testing, reporting sensitivity, specificity, precision, F1 score, and overall accuracy.</p> Results <p>Analysis of Gray Level Co-occurrence Matrix (GLCM) features revealed significant textural disparities between normal and OSCC. Specifically, OSCC images showed trends of finer texture (lower autocorrelation), higher local intensity variation, increased complexity (entropy), greater intensity disparities (dissimilarity), and reduced uniformity (lower energy, maximum probability, and homogeneity) compared to normal counterparts. Leveraging these distinct textural signatures, the developed machine learning models exhibited exceptional classification capabilities, frequently achieving 100% sensitivity, specificity, precision, F1 score, and overall accuracy across all tasks during both validation and testing of models. This high performance highlights the reliability of the proposed approach for differentiating OSCC.</p> Conclusions <p>This study demonstrated that statistical image features effectively capture distinct textural differences between normal and OSCC tissues. The developed machine learning models showed outstanding classification performance, demonstrating their strong potential for accurate, automated OSCC diagnosis. This near-perfect accuracy could revolutionize diagnostic workflows, minimizing subjective interpretations and enhancing overall efficiency.</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

Texture-based image feature analysis for the classification of oral squamous cell carcinoma using machine learning approach

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月27日

查看全文
  • Salma Bahaa,
  • Omneya Ramadan,
  • Zeinab Elsayed Darwish,
  • Eslam EL-Fiky,
  • Mai M. Saleh

摘要

Background

Artificial intelligence is being developed as a vital tool to improve the accuracy and efficiency of oral cancer diagnosis. Oral squamous cell carcinoma is often detected at advanced stages, leading to poor patient outcomes, and traditional diagnostic methods can be subjective and physically demanding for pathologists. By comparing different Artificial intelligence models, including machine learning and deep learning, this study aims to create a more standardized and rapid diagnostic aid that can support pathologists and increase patient survival rates.

Methods

A total of 115 cases (49 OSCC, 66 normal) were captured at 100x and 400x magnifications. Twelve Gray Level Co-occurrence Matrix (GLCM) features were extracted. Machine learning models, including Fine Tree, Medium Tree, Linear Discriminant, Linear Support Vector Machine, K-Nearest Neighbors, and Neural Network (MLP) model were developed. Model performance was evaluated across four distinct classification tasks, assessing the contribution of histopathological images alone and in fusion with demographic/clinical data. Statistical validation involved 5-fold cross-validation on the training set and independent testing, reporting sensitivity, specificity, precision, F1 score, and overall accuracy.

Results

Analysis of Gray Level Co-occurrence Matrix (GLCM) features revealed significant textural disparities between normal and OSCC. Specifically, OSCC images showed trends of finer texture (lower autocorrelation), higher local intensity variation, increased complexity (entropy), greater intensity disparities (dissimilarity), and reduced uniformity (lower energy, maximum probability, and homogeneity) compared to normal counterparts. Leveraging these distinct textural signatures, the developed machine learning models exhibited exceptional classification capabilities, frequently achieving 100% sensitivity, specificity, precision, F1 score, and overall accuracy across all tasks during both validation and testing of models. This high performance highlights the reliability of the proposed approach for differentiating OSCC.

Conclusions

This study demonstrated that statistical image features effectively capture distinct textural differences between normal and OSCC tissues. The developed machine learning models showed outstanding classification performance, demonstrating their strong potential for accurate, automated OSCC diagnosis. This near-perfect accuracy could revolutionize diagnostic workflows, minimizing subjective interpretations and enhancing overall efficiency.