Enhanced Knee Osteoarthritis Grading: Transfer Learning with Pre-Trained CNN’s For Swift Diagnosis

DOI

10.2991/978-94-6463-471-6_2How to use a DOI?

Keywords

Knee Osteoarthritis; Kellgren-Lawrence; Artificial Intelligence; Deep learning; Computer aided diagnosis; Rectified Linear Unit (ReLU); Partial fine-tuning

Abstract

Knee osteoarthritis (KOA) is a common long-lasting ailment characterized by joint degradation, impacting millions worldwide and posing significant challenges in early detection and management. Radiographic assessment, particularly through the Kellgren-Lawrence (KL) grading scheme, serves as a cornerstone for diagnosis and disease monitoring. However, the subjective nature of visual examination, coupled with the expertise-dependent interpretation, often leads to variations in grading accuracy and delays in intervention. This study introduces an enhanced Computer-Aided Diagnostic (CAD) system framework leveraging Adaptive Pretrained Convolutional Neural Networks (CNNs) for rapid and precise classification of KOA severity based on KL grading By learning on large-scale datasets, transfer learning is utilized to create robust representations, which helps to mitigate the difficulties brought on by the scarcity of medical data. The performance of these transfer learning is enhanced by introducing adaptability to refining through partial fine-tuning of the layers to accommodate the specific nuances of KOA grading. The effectivity of the proposed enhancement is evaluated by using an extensive experimental setup in which eight pretrained CNN models, namely NasNetLarge, InceptionV3, DenseNet169, ResNet152V2, NasNetMobile, ResNet50V2, ResNet101V2 and InceptionResNetV2 are used for KOA using KL grading in clinical scenario. The proposed enhanced approach is applied on Kaggle X-ray data sets repository. The method demonstrates promising results in accurately classifying KOA severity, offering a significant advancement towards early detection and personalized management strategies. The results show that NasNetLarge is the most efficient model, with a validation accuracy of 98.8%, training accuracy of 98.3% and precision of 99.7%. By combining pretrained CNN models and transfer learning with partial fine-tuning, the proposed framework facilitates rapid and cost-effective diagnosis, contributing to early diagnosis and help to reduce progression.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - V. Vijaya Kishore
AU  - K. Sahithi
AU  - K. Siva Jayanth Reddy
AU  - K. Akash
AU  - K. Sai Jyothy
AU  - Sreekanth Yalavarthi
PY  - 2024
DA  - 2024/07/30
TI  - Enhanced Knee Osteoarthritis Grading: Transfer Learning with Pre-Trained CNN’s For Swift Diagnosis
BT  - Proceedings of the International Conference on Computational Innovations and Emerging Trends (ICCIET- 2024)
PB  - Atlantis Press
SP  - 4
EP  - 20
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-471-6_2
DO  - 10.2991/978-94-6463-471-6_2
ID  - Kishore2024
ER  -