Scholarlink

Nephrolithiasis, or kidney stones, is a common painful condition affecting millions of people around the world and has significant health and economic implications. Traditional methods of diagnosis are X-rays, ultrasound, and CT scans. These methods have several limitations: radiation exposure, operator dependency, and inconsistent results. This study covers the application of a CNN-based convolution neural network method for the classification of kidney stones on images taken using a medical imaging approach. Using a data set developed using CT and scans to enhance the model based on robust deep learning methods that potentially lead to an autonomous detection process. The performance of the CNN model was remarkable: precision, 99.33%, with a perfect recall at 100%, while its low loss value is at 0.227. It had an impressive generalization performance, so much so that precision is at an incredibly high number that will make the tool reliable for real use in the diagnostic process. Therefore, this study will open up ways for AI-based diagnostic tools to enhance early detection and reduce healthcare burdens while improving patient outcomes, paving the way for more accessible and reliable kidney stone diagnostics.

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

CNN-Based Deep Learning for Automated Kidney Stone Classification in Medical Imaging (CT Scan)

  • 会议论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年

查看全文
  • Vishesh Tanwar,
  • Bhisham Sharma,
  • Imed Ben Dhaou

摘要

Nephrolithiasis, or kidney stones, is a common painful condition affecting millions of people around the world and has significant health and economic implications. Traditional methods of diagnosis are X-rays, ultrasound, and CT scans. These methods have several limitations: radiation exposure, operator dependency, and inconsistent results. This study covers the application of a CNN-based convolution neural network method for the classification of kidney stones on images taken using a medical imaging approach. Using a data set developed using CT and scans to enhance the model based on robust deep learning methods that potentially lead to an autonomous detection process. The performance of the CNN model was remarkable: precision, 99.33%, with a perfect recall at 100%, while its low loss value is at 0.227. It had an impressive generalization performance, so much so that precision is at an incredibly high number that will make the tool reliable for real use in the diagnostic process. Therefore, this study will open up ways for AI-based diagnostic tools to enhance early detection and reduce healthcare burdens while improving patient outcomes, paving the way for more accessible and reliable kidney stone diagnostics.