Scholarlink

Osteoarthritis (OA) of the knee, often referred to as degenerative joint disease, results from the degradation of articular cartilage and the cumulative wear and tear of the joint over time. In advanced stages, total knee replacement (TKR) surgery becomes necessary, which is both invasive and costly, significantly affecting the patient’s quality of life. This Manuscript aims to leverage raw radiographic data to forecast the structural evolution of OA, potentially offering non-invasive, predictive insights that could improve patient outcomes. A Deep Convolutional Neural Network (CNN) is adopted for predicting the severity of OA. A comparative analysis on Five Machine learning models namely- ResNet101, MobileNetV2, InceptionResNetV2, SE-ResNet50 is done. In addition, a Hybrid architecture is implemented by modifying the SE-ResNet50 architecture. This model aimed to extract subtle features from both image and structured data, enhancing its predictive capabilities. The dataset for this analysis is sourced from the Osteoarthritis Initiative (OAI). A dataset of 9793 images is considered by categorizing them into 5 classes. This is categorized based on severity levels with Kellgren–Lawrence (KL) grading system: 0 (healthy), 1 (doubtful), 2 (minimal), 3 (moderate), and 4 (severe). The dataset is sourced from the Osteoarthritis Initiative (OAI). Through the integration of Cycle GAN techniques for data augmentation, the study aims to enhance dataset diversity and resilience. Emphasizing inclusivity, a user-friendly web interface is developed to ensure seamless interaction for diverse users. From the comparative analysis the hybrid model has outperformed with a testing accuracy of 87.8%.

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 Deep Belief Network Architecture for Knee Osteoarthritis Prediction - Data Deployment

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Kakarla Deepti,
  • S. A. Deepthi,
  • Maithri Jajala,
  • Singaraju Sreya,
  • Komirelly Anishka Reddy

摘要

Osteoarthritis (OA) of the knee, often referred to as degenerative joint disease, results from the degradation of articular cartilage and the cumulative wear and tear of the joint over time. In advanced stages, total knee replacement (TKR) surgery becomes necessary, which is both invasive and costly, significantly affecting the patient’s quality of life. This Manuscript aims to leverage raw radiographic data to forecast the structural evolution of OA, potentially offering non-invasive, predictive insights that could improve patient outcomes. A Deep Convolutional Neural Network (CNN) is adopted for predicting the severity of OA. A comparative analysis on Five Machine learning models namely- ResNet101, MobileNetV2, InceptionResNetV2, SE-ResNet50 is done. In addition, a Hybrid architecture is implemented by modifying the SE-ResNet50 architecture. This model aimed to extract subtle features from both image and structured data, enhancing its predictive capabilities. The dataset for this analysis is sourced from the Osteoarthritis Initiative (OAI). A dataset of 9793 images is considered by categorizing them into 5 classes. This is categorized based on severity levels with Kellgren–Lawrence (KL) grading system: 0 (healthy), 1 (doubtful), 2 (minimal), 3 (moderate), and 4 (severe). The dataset is sourced from the Osteoarthritis Initiative (OAI). Through the integration of Cycle GAN techniques for data augmentation, the study aims to enhance dataset diversity and resilience. Emphasizing inclusivity, a user-friendly web interface is developed to ensure seamless interaction for diverse users. From the comparative analysis the hybrid model has outperformed with a testing accuracy of 87.8%.