Scholarlink

<p>Early and accurate detection of prostate cancer plays a vital role in improving treatment outcomes and reducing mortality among men worldwide. Low specificity and the possibility of overdiagnosis are two common drawbacks of traditional diagnostic techniques including transrectal ultrasound-guided biopsy and prostate-specific antigen (PSA) testing. Deep learning (DL), a subset of artificial intelligence (AI), has become a potent technique in recent years for improving prostate cancer detection, providing increased precision, effectiveness, and repeatability. With an emphasis on magnetic resonance imaging (MRI)-based methods, this systematic review investigates the state of DL applications in prostate cancer analysis. Rich anatomical and functional detail is provided by MRI, particularly in its multi-parametric version (mpMRI), which is crucial for accurate tumor localization and risk assessment. The review compares the performance of transformer-based designs with convolutional neural networks (CNNs) in classifying DL models across important tasks like classification, segmentation, lesion detection, and progression forecasting. Additionally, it emphasizes preprocessing approaches, multimodal data integration strategies, and publically accessible datasets. There is a critical discussion of the factors that hinder clinical adoption, such as generalizability, regulatory restrictions, and model interpretability. This study offers a thorough resource for academics and clinicians looking to develop AI-driven prostate cancer diagnosis through MRI-based deep learning by combining results from more than 180 investigations.</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

A review of deep learning methods for the prediction of prostate cancer

  • 期刊论文

  • 发表日期：

    2025年11月28日

查看全文
  • Rasool Al-Gburi,
  • Raid Gaib,
  • Saif M. B. Al-Sabti,
  • Muhammed E. Tharwat,
  • Ali Mustafa,
  • Ali M. Elhendy

摘要

Early and accurate detection of prostate cancer plays a vital role in improving treatment outcomes and reducing mortality among men worldwide. Low specificity and the possibility of overdiagnosis are two common drawbacks of traditional diagnostic techniques including transrectal ultrasound-guided biopsy and prostate-specific antigen (PSA) testing. Deep learning (DL), a subset of artificial intelligence (AI), has become a potent technique in recent years for improving prostate cancer detection, providing increased precision, effectiveness, and repeatability. With an emphasis on magnetic resonance imaging (MRI)-based methods, this systematic review investigates the state of DL applications in prostate cancer analysis. Rich anatomical and functional detail is provided by MRI, particularly in its multi-parametric version (mpMRI), which is crucial for accurate tumor localization and risk assessment. The review compares the performance of transformer-based designs with convolutional neural networks (CNNs) in classifying DL models across important tasks like classification, segmentation, lesion detection, and progression forecasting. Additionally, it emphasizes preprocessing approaches, multimodal data integration strategies, and publically accessible datasets. There is a critical discussion of the factors that hinder clinical adoption, such as generalizability, regulatory restrictions, and model interpretability. This study offers a thorough resource for academics and clinicians looking to develop AI-driven prostate cancer diagnosis through MRI-based deep learning by combining results from more than 180 investigations.