A fully automatic target detection and quantification strategy based on object detection convolutional neural network...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Clinical diagnosis of osteoarthritis (OA) relies heavily on X-ray imaging, generating vast amounts of data.
• Efficient and automated image analysis methods are crucial for managing this data explosion in healthcare.

Purpose of the Study:

• To develop a novel deep learning model for simultaneous knee joint localization and radiographic osteoarthritis (OA) severity quantification.
• To leverage the YOLO version 3 (YOLOv3) algorithm for accurate and efficient OA assessment.

Main Methods:

• A unified Convolutional Neural Network (CNN) model was designed using the YOLOv3 framework.
• The model integrates knee joint detection and OA severity grading capabilities.
• The model was trained and validated on public and clinical X-ray datasets.

Main Results:

• The YOLOv3-based model demonstrated desirable accuracy in grading knee OA severity.
• Significant improvements were observed in precision, recall, F1 score, and diagnostic accuracy.
• The automated system processes images in just 40 ms, enabling rapid clinical decisions.

Conclusions:

• The proposed YOLOv3 strategy offers a convenient and efficient solution for automatic knee OA diagnosis from X-ray images.
• This automated approach supports quick clinical decision-making and streamlines daily diagnostic workflows.