Scholarlink

<p>Kidney stone disease is a widespread urological condition affecting approximately 10% of the global population and demands prompt and accurate diagnosis to avoid severe complications. This study proposes a deep learning-based computer-aided diagnostic (CADx) framework for the detection of kidney stones using axial CT images. A curated dataset comprising 3,364 annotated CT scans and 35,457 augmented images generated via twelve targeted augmentation strategies was collected from three medical centers in the Kurdistan Region of Iraq: Mercy Medical City, Faruq Medical City, and Rania Medical City. A convolutional neural network (CNN) was designed and trained for binary classification (“stone” vs. “non-stone”), leveraging convolutional layers, ReLU activation, max pooling, and fully connected layers to extract and interpret high-level features. The model was evaluated using a patient-wise data split to prevent leakage and ensure generalizability. It achieved a classification accuracy of 98.80%, with consistently strong performance across precision, recall, sensitivity, specificity, and F1-score metrics. These results demonstrate the effectiveness of the model in assisting radiologists by reducing manual diagnostic workload, minimizing human error, and improving clinical workflow. The outcomes of this research highlight the growing potential of artificial intelligence in radiology and underscore the critical role of high-quality datasets in enabling robust and reliable diagnostic systems.</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 high-accurate model for kidney stone detection in CT images using advanced deep learning techniques

  • 期刊论文

  • 发表日期：

    2026年6月30日

查看全文
  • Peshraw Ahmed Abdalla,
  • Bander Sidiq Mahmood,
  • Nawzad Rasul Hama

摘要

Kidney stone disease is a widespread urological condition affecting approximately 10% of the global population and demands prompt and accurate diagnosis to avoid severe complications. This study proposes a deep learning-based computer-aided diagnostic (CADx) framework for the detection of kidney stones using axial CT images. A curated dataset comprising 3,364 annotated CT scans and 35,457 augmented images generated via twelve targeted augmentation strategies was collected from three medical centers in the Kurdistan Region of Iraq: Mercy Medical City, Faruq Medical City, and Rania Medical City. A convolutional neural network (CNN) was designed and trained for binary classification (“stone” vs. “non-stone”), leveraging convolutional layers, ReLU activation, max pooling, and fully connected layers to extract and interpret high-level features. The model was evaluated using a patient-wise data split to prevent leakage and ensure generalizability. It achieved a classification accuracy of 98.80%, with consistently strong performance across precision, recall, sensitivity, specificity, and F1-score metrics. These results demonstrate the effectiveness of the model in assisting radiologists by reducing manual diagnostic workload, minimizing human error, and improving clinical workflow. The outcomes of this research highlight the growing potential of artificial intelligence in radiology and underscore the critical role of high-quality datasets in enabling robust and reliable diagnostic systems.