Scholarlink

<p>Idiopathic scoliosis can significantly negatively affect the individual’s quality of life and cause physical and psychological problems. The first step in the treatment process of the disease is to determine the Cobb angle. Alternative treatment methods such as brace treatment, physical therapy and exercise, surgical intervention, only follow-up or acupuncture can be applied to the patient according to the Cobb angle. Therefore, the Cobb angle plays a critical role in determining the treatment method to be applied to the patient. In this study, a hybrid ViT model that can perform multi-class scoliosis classification according to the Cobb angle from X-ray images was proposed. The proposed model uses ViT Base Patch16 and ViT Base Patch32 models to obtain both micro- and macro-scale structural information in X-ray images. Classification was performed with machine learning algorithms by combining the feature maps obtained from both models. 93.80% accuracy and 99.63% AUC values were obtained with the logistic regression classifier.</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

Multi-class idiopathic scoliosis detection based on Cobb angle using a hybrid ViT model

  • 期刊论文

  • 发表日期：

    2025年9月26日

查看全文
  • Nevzat Yeşilmen,
  • Merve Parlak Baydoğan,
  • Çağla Danacı,
  • Seda Arslan Tuncer,
  • Ahmet Çınar,
  • Taner Tuncer

摘要

Idiopathic scoliosis can significantly negatively affect the individual’s quality of life and cause physical and psychological problems. The first step in the treatment process of the disease is to determine the Cobb angle. Alternative treatment methods such as brace treatment, physical therapy and exercise, surgical intervention, only follow-up or acupuncture can be applied to the patient according to the Cobb angle. Therefore, the Cobb angle plays a critical role in determining the treatment method to be applied to the patient. In this study, a hybrid ViT model that can perform multi-class scoliosis classification according to the Cobb angle from X-ray images was proposed. The proposed model uses ViT Base Patch16 and ViT Base Patch32 models to obtain both micro- and macro-scale structural information in X-ray images. Classification was performed with machine learning algorithms by combining the feature maps obtained from both models. 93.80% accuracy and 99.63% AUC values were obtained with the logistic regression classifier.