Graph-Based Motion Artifacts Detection Method from Head Computed Tomography Images.

Area of Science:

• Medical Imaging
• Network Science
• Artificial Intelligence

Background:

• Computed tomography (CT) is crucial for diagnosis, but motion artifacts degrade image quality and reduce accuracy.
• Convolutional Neural Networks (CNNs) excel in medical imaging but are computationally intensive and lack interpretability.
• Existing methods struggle with the complexity and resource demands of deep learning for artifact detection.

Purpose of the Study:

• To develop an interpretable and efficient method for detecting motion artifacts in head CT images.
• To address the limitations of CNNs in terms of memory usage, computational resources, and lack of transparency.
• To improve the diagnostic accuracy of CT scans by accurately identifying motion artifacts.

Main Methods:

• A novel motion artifact detection method based on complex networks (MADM-CN) was proposed.
• CT images were represented as complex networks, and artifacts were detected by classifying these networks.
• Network topological characteristics, including average degree and clustering coefficient, were used for classification.

Main Results:

• The MADM-CN method demonstrated superior performance compared to conventional machine learning and deep learning approaches.
• The proposed method achieved up to 98% accuracy and 97% sensitivity in detecting motion artifacts.
• The graph-based approach provided an interpretable alternative for artifact detection.

Conclusions:

• The MADM-CN method offers an effective and interpretable solution for motion artifact detection in head CT images.
• This approach overcomes the 'black-box' nature and resource intensiveness of CNNs.
• The findings suggest a promising direction for improving the reliability of CT-based diagnoses.