Scholarlink

One type of bacterium known as Mycobacterium tuberculosis is the cause of tuberculosis (TB), an endemic disease. The lungs and other areas of the body are the main targets. When an infected person coughs, sneezes, or spits, the disease can spread through the air and water droplets. Convolutional Neural Networks (CNNs), a deep learning method, are suggested in this extension for the identification of tuberculosis (TB) in chest X-ray pictures. In order to capture the important aspects of the images, such as size, shape, texture, and intensity, the model is built on the UNET architecture. Based on a sizable dataset of about 4300 chest X-ray images, the suggested model’s performance is assessed and contrasted with multiple architectures, including Resnet50, InceptionResnetV2, MobileNet, EfficientB3, and U-Net. According to the results, the recommended model outperforms the others in terms of F1 score (91%), accuracy (98%), precision (96%), recall (95%), specificity (99%), and recall (95%). As seen by the evaluation criteria of specificity, accuracy, and precision, the proposed UNET model outperforms previous state-of-the-art models, indicating its potential as an early tuberculosis detection tool.

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

Tuberculosis Detection in Chest X-rays Using Deep Learning Algorithms with Segmentation and Data Augmentation Techniques

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jarapla Suharthi,
  • Deshapathi Sujitha,
  • Akarapu Varun Mithra,
  • Kadiyala Ragodaya Deepthi

摘要

One type of bacterium known as Mycobacterium tuberculosis is the cause of tuberculosis (TB), an endemic disease. The lungs and other areas of the body are the main targets. When an infected person coughs, sneezes, or spits, the disease can spread through the air and water droplets. Convolutional Neural Networks (CNNs), a deep learning method, are suggested in this extension for the identification of tuberculosis (TB) in chest X-ray pictures. In order to capture the important aspects of the images, such as size, shape, texture, and intensity, the model is built on the UNET architecture. Based on a sizable dataset of about 4300 chest X-ray images, the suggested model’s performance is assessed and contrasted with multiple architectures, including Resnet50, InceptionResnetV2, MobileNet, EfficientB3, and U-Net. According to the results, the recommended model outperforms the others in terms of F1 score (91%), accuracy (98%), precision (96%), recall (95%), specificity (99%), and recall (95%). As seen by the evaluation criteria of specificity, accuracy, and precision, the proposed UNET model outperforms previous state-of-the-art models, indicating its potential as an early tuberculosis detection tool.