The Knee · 2026
Knee osteoarthritis classification using Optimized Granger Causality Inspired Graph Neural Network with Accelerated Model-agnostic Explanations.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Ganesh Kumar M, Bharathi L, Senthil M, Koteswara Rao P, Rajya Lakshmi G
Аннотация
BACKGROUND: Knee osteoarthritis (KOA) is a common, degenerative joint condition where cartilage wears away, causing bone-on-bone friction, leading to pain, stiffness, swelling, and reduced mobility, often with a cracking sound (crepitus). KOA affects the knee joint as a whole, and makes it difficult for the knee to move normally. In this paper, KOA classification using Optimized Granger Causality Inspired Graph Neural Network with Accelerated Model-agnostic Explanations and Secretary Bird Optimization Algorithm (KOAC-GCIGNN-AcME-SBOA) is proposed.
METHODS: The input images were collected from the Osteoarthritis Initiative (OAI) database. The images were then fed into the pre-processing stage with the help of Multi-Window Savitzky-Golay Filter (MWSGF) for artifact removal, resizing, contrast handling and normalization of the image. The pre-processed images were then fed into the feature extraction stage. The feature extraction was performed by Feature Affine Residual Network (FA-ResNet) to extract features such as mean, median, standard deviation, kurtosis and skewness. Finally, the extracted features were fed into GCIGNN-AcME for classifying KOA detection severity depending on four grades: Grade 0 (healthy), Grade 1 (doubtful), Grade 2 (minimal), Grade 3 (moderate) and Grade 4 (severe). Finally, SBOA was proposed to optimize the weight parameter of GCIGNN-AcME for KOA detection. Metrics, such as accuracy, precision, f1-score, sensitivity, specificity, receiver operating characteristic, and computational time were evaluated.
RESULTS: The KOAC-GCIGNN-AcME-SBOA attained 26.36%, 20.69%, 30.29% higher accuracy, 19.12%, 28.32%, 27.84% higher precision, 12.04%, 13.45%, 22.80% higher sensitivity compared with the existing methods: fully automated, fine-tuned deep learning method for the study of KOA progression (FAFT-CNN-KOD), KOA severity classification method with ordinal regression module (KAC-DNN-ORM), and KOA detection and classification method utilizing X-rays (KOD-CNN-XR), respectively.
CONCLUSION: The proposed KOAC-GCIGNN-AcME-SBOA was successfully implemented.
METHODS: The input images were collected from the Osteoarthritis Initiative (OAI) database. The images were then fed into the pre-processing stage with the help of Multi-Window Savitzky-Golay Filter (MWSGF) for artifact removal, resizing, contrast handling and normalization of the image. The pre-processed images were then fed into the feature extraction stage. The feature extraction was performed by Feature Affine Residual Network (FA-ResNet) to extract features such as mean, median, standard deviation, kurtosis and skewness. Finally, the extracted features were fed into GCIGNN-AcME for classifying KOA detection severity depending on four grades: Grade 0 (healthy), Grade 1 (doubtful), Grade 2 (minimal), Grade 3 (moderate) and Grade 4 (severe). Finally, SBOA was proposed to optimize the weight parameter of GCIGNN-AcME for KOA detection. Metrics, such as accuracy, precision, f1-score, sensitivity, specificity, receiver operating characteristic, and computational time were evaluated.
RESULTS: The KOAC-GCIGNN-AcME-SBOA attained 26.36%, 20.69%, 30.29% higher accuracy, 19.12%, 28.32%, 27.84% higher precision, 12.04%, 13.45%, 22.80% higher sensitivity compared with the existing methods: fully automated, fine-tuned deep learning method for the study of KOA progression (FAFT-CNN-KOD), KOA severity classification method with ordinal regression module (KAC-DNN-ORM), and KOA detection and classification method utilizing X-rays (KOD-CNN-XR), respectively.
CONCLUSION: The proposed KOAC-GCIGNN-AcME-SBOA was successfully implemented.