Scholarlink

<p>Prostate cancer (PCa) is one of the most prevalent malignancies in males, demanding advanced imaging techniques for early detection, precise staging, and the formulation of treatment strategies. Single Photon Emission Computed Tomography (SPECT) has emerged as a pivotal imaging modality, particularly for PCa metabolic and functional evaluation. Unlike structural imaging methods such as magnetic resonance imaging (MRI), SPECT provides critical insights into the biological activity of tumors, making it indispensable for assessing cancer aggressiveness and monitoring therapeutic responses. However, challenges such as limited spatial resolution, noise artifacts, and restricted quantification capabilities have historically hindered its widespread application. This study focuses on the application of SPECT/CT and FastMRI datasets for the detection of prostate cancer. The proposed method employs principal component analysis (PCA) for feature extraction and dimensionality reduction from SPECT/CT and FastMRI datasets. These extracted features are then utilized in a neural network (NN) to classify the Gleason score. The model achieved a performance accuracy of 79% on a limited dataset comprising 87 SPECT/CT samples. This represents a 41% improvement in accuracy compared to previous models relying on radiomic features for extraction. The results underscore the model’s robustness, even when faced with small, imbalanced, and low-quality datasets.</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

Enhancing prostate cancer diagnosis using 99mTc-PSMA SPECT: PCA-based feature extraction and neural network classification for Gleason score analysis

  • 期刊论文

  • 发表日期：

    2026年3月3日

查看全文
  • Masoumeh Dorri Giv,
  • Samer Kais Jameel,
  • Jafar Majidpour,
  • Sayna Jamaati,
  • Hossein Arabi

摘要

Prostate cancer (PCa) is one of the most prevalent malignancies in males, demanding advanced imaging techniques for early detection, precise staging, and the formulation of treatment strategies. Single Photon Emission Computed Tomography (SPECT) has emerged as a pivotal imaging modality, particularly for PCa metabolic and functional evaluation. Unlike structural imaging methods such as magnetic resonance imaging (MRI), SPECT provides critical insights into the biological activity of tumors, making it indispensable for assessing cancer aggressiveness and monitoring therapeutic responses. However, challenges such as limited spatial resolution, noise artifacts, and restricted quantification capabilities have historically hindered its widespread application. This study focuses on the application of SPECT/CT and FastMRI datasets for the detection of prostate cancer. The proposed method employs principal component analysis (PCA) for feature extraction and dimensionality reduction from SPECT/CT and FastMRI datasets. These extracted features are then utilized in a neural network (NN) to classify the Gleason score. The model achieved a performance accuracy of 79% on a limited dataset comprising 87 SPECT/CT samples. This represents a 41% improvement in accuracy compared to previous models relying on radiomic features for extraction. The results underscore the model’s robustness, even when faced with small, imbalanced, and low-quality datasets.