Ensemble CycleGAN for Retrospective Rigid Motion Correction in MRI.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Image Processing

Background:

• Patient motion during Magnetic Resonance Imaging (MRI) scans introduces artifacts, compromising diagnostic accuracy.
• Existing motion correction methods often struggle with diverse artifact types, necessitating advanced solutions.

Purpose of the Study:

• To develop and evaluate a novel Retrospective Motion Correction (RMC) technique for MRI using an Ensemble Cycle-consistent Generative Adversarial Network (CycleGAN).
• To enhance the robustness and effectiveness of motion artifact correction by combining multiple specialized CycleGAN models.

Main Methods:

• Development of an Ensemble CycleGAN incorporating three distinct CycleGAN models, each trained to address specific motion artifact types.
• Generation of motion-corrupted MRI data using a comprehensive Motion Simulation Tool for training and validation.
• Quantitative evaluation of the proposed method using Structural Similarity Index Measure (SSIM) and Normalized Mean Squared Error (NMSE) on brain MRI slices.

Main Results:

• The Ensemble CycleGAN approach demonstrated superior performance in correcting motion artifacts compared to single CycleGAN models.
• Significant improvements in image quality and diagnostic potential were observed following motion correction.
• The method effectively leveraged the strengths of individual models to address a broader range of motion-induced artifacts.

Conclusions:

• The proposed Ensemble CycleGAN method offers a powerful and effective solution for retrospective motion correction in MRI.
• This technique has the potential to improve the reliability and accuracy of MRI diagnostics in clinical settings.
• Further research can explore the application of this ensemble approach to other imaging modalities and artifact types.