DIKOApp: An AI-Based Diagnostic System for Knee Osteoarthritis

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Knee osteoarthritis diagnosis is complex and challenging.
• Existing AI methods face difficulties in real-world medical image analysis.
• Unique biological characteristics in specific populations, like Vietnamese individuals, limit the applicability of prior studies.

Purpose of the Study:

• To introduce DIKOApp, an automatic diagnostic application for knee osteoarthritis.
• To address challenges in AI-based knee osteoarthritis diagnosis, particularly for the Vietnamese population.
• To leverage a novel DIKO framework combining medical knowledge and computer vision.

Main Methods:

• Developed DIKOApp based on the DIKO framework, utilizing a two-stage approach.
• Trained the model on a dataset specifically curated for the Vietnamese population.
• Employed efficient data sampling and augmentation techniques for real-world medical images.

Main Results:

• The DIKO model achieved an accuracy of 89.34% on a real-world dataset of Vietnamese individuals.
• An F1-score of 0.88 was obtained, indicating strong diagnostic performance.
• The application demonstrated practical potential for clinical use.

Conclusions:

• DIKOApp provides an effective and accurate AI-powered tool for knee osteoarthritis diagnosis.
• The framework shows promise for improving diagnostic accuracy and reducing time for healthcare providers.
• This application has significant potential to enhance patient care and outcomes.