[Reduction of Motion Artifacts in Liver MRI Using Deep Learning with High-pass Filtering]

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Image Processing

Context:

• Motion artifacts are a common problem in liver Magnetic Resonance Imaging (MRI).
• These artifacts can degrade image quality and potentially affect diagnostic accuracy.
• Developing effective methods to reduce motion artifacts is crucial for reliable liver MRI interpretation.

Purpose:

• To evaluate the efficacy of a deep learning model incorporating high-pass filtering for reducing motion artifacts in liver MRI.
• To assess the impact of this deep learning approach on image sharpness and diagnostic quality.

Summary:

• A deep learning model was trained using simulated motion artifact images (SMAIs) derived from non-artifact images (NAIs).
• The model generated motion artifact reduction images (MARIs) which were quantitatively assessed using Structural Similarity Index Measure (SSIM) and contrast ratio (CR).
• Visual assessment compared MARIs against motion artifact images (MAIs), evaluating artifact reduction and image sharpness.

Impact:

• The deep learning model successfully reduced motion artifacts in liver MRI.
• Image sharpness was maintained or improved in the processed images (MARIs) compared to artifact-laden images (MAIs).
• This technique offers a promising solution for enhancing the quality of liver MRI, aiding in more accurate diagnoses.