Joint Neural Network for Fast Retrospective Rigid Motion Correction of Accelerated Segmented Multislice MRI.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Motion artifacts are a major challenge in Magnetic Resonance Imaging (MRI), degrading image quality and potentially leading to misdiagnosis.
• Existing motion correction techniques often require additional hardware or complex tracking systems, limiting their applicability.
• Retrospective correction methods aim to fix motion artifacts after the scan, but often struggle with speed and accuracy.

Purpose of the Study:

• To introduce a novel deep learning approach for fast retrospective intraslice rigid motion correction in segmented multislice MRI.
• To develop a method that does not require auxiliary information on subject head motion during the scan.
• To achieve high-quality image reconstructions that are both accurate and computationally efficient.

Main Methods:

• A neural network architecture combining frequency and image space convolutions was designed for motion correction.
• The deep learning model processes acquired MRI data to reconstruct motion-free images.
• The method was evaluated for its performance in correcting intraslice rigid motion in segmented multislice MRI scans.

Main Results:

• The proposed deep learning method achieved high-quality image reconstructions.
• The reconstruction procedure was found to be more accurate compared to existing methods.
• The method demonstrated a significant speed improvement, being an order of magnitude faster than GRAPPA.

Conclusions:

• Deep learning offers a powerful tool for fast and accurate retrospective motion correction in MRI.
• The developed method provides a promising solution for challenging motion scenarios in MRI, including substantial and unpredictable head movements.
• This work represents a significant advancement towards real-time motion correction in various MRI applications.