Scholarlink

<p>Breast cancer remains a leading cause of cancer-related mortality among women worldwide, underscoring the urgency for early and accurate detection strategies. While traditional research has primarily emphasized mammography and binary classification tasks, emerging advances in artificial intelligence (AI) now enable integration of multimodal data, multi-class severity assessment, and enhanced interpretability. In this review, we synthesize the existing literature on AI-driven approaches for breast cancer detection and risk prediction, and complement it with an empirical comparative evaluation across clinical data, ultrasound, and computed tomography imaging. Both traditional machine learning classifiers and deep learning architectures are examined, with ensemble and transfer learning methods demonstrating superior diagnostic accuracy. Additionally, text-based models provide valuable granularity in severity classification, and explainable AI (XAI) techniques such as Grad-CAM enhance transparency by generating interpretable visual insights. Together, these findings highlight the potential of hybrid and explainable AI systems to improve precision, reliability, and clinical applicability in breast cancer management. By integrating a broad review with comparative analysis, this work provides an evidence-based perspective on current advances, limitations, and future directions toward personalized and trustworthy AI solutions in oncology.</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

Early Detection and Personalized Risk Assessment of Breast Cancer: An Integrative Review and Comparative Evaluation of AI Models Using Clinical and Imaging Data

  • 期刊论文

  • 发表日期：

    2025年12月9日

查看全文
  • Kruti Kikani,
  • Mitali Desai,
  • Binita Desai

摘要

Breast cancer remains a leading cause of cancer-related mortality among women worldwide, underscoring the urgency for early and accurate detection strategies. While traditional research has primarily emphasized mammography and binary classification tasks, emerging advances in artificial intelligence (AI) now enable integration of multimodal data, multi-class severity assessment, and enhanced interpretability. In this review, we synthesize the existing literature on AI-driven approaches for breast cancer detection and risk prediction, and complement it with an empirical comparative evaluation across clinical data, ultrasound, and computed tomography imaging. Both traditional machine learning classifiers and deep learning architectures are examined, with ensemble and transfer learning methods demonstrating superior diagnostic accuracy. Additionally, text-based models provide valuable granularity in severity classification, and explainable AI (XAI) techniques such as Grad-CAM enhance transparency by generating interpretable visual insights. Together, these findings highlight the potential of hybrid and explainable AI systems to improve precision, reliability, and clinical applicability in breast cancer management. By integrating a broad review with comparative analysis, this work provides an evidence-based perspective on current advances, limitations, and future directions toward personalized and trustworthy AI solutions in oncology.