Image registration in dynamic renal MRI-current status and prospects

Area of Science:

• Medical Imaging
• Nephrology
• Biomedical Engineering

Background:

• Magnetic resonance imaging (MRI) is crucial for diagnosing chronic kidney diseases (CKD).
• Dynamic contrast-enhanced (DCE-) MRI and arterial spin labeling (ASL) assess hemodynamic parameters like renal blood flow and glomerular filtration rate (GFR).
• Motion artifacts from physiological processes significantly impede accurate analysis of time-resolved renal MRI data.

Purpose of the Study:

• To review existing literature on renal image registration techniques for mitigating motion artifacts in time-resolved renal MRI.
• To discuss the limitations of current techniques hindering clinical translation.
• To explore traditional and emerging registration methods, including deep learning approaches.

Main Methods:

• Categorization of renal image registration strategies into acquisition techniques, post-processing methods, or combined approaches.
• Analysis of traditional registration components: transformation, criterion function, and search types.
• Inclusion of emerging deep learning-based registration technologies.

Main Results:

• Various strategies exist to address motion artifacts in renal MRI, including image registration.
• Deep learning-based methods show promise for faster and more accurate registrations.
• Despite progress, a significant gap remains between proposed techniques and clinical practice.

Conclusions:

• Standardized evaluation protocols are critically needed for renal MRI image registration.
• Overcoming motion artifacts is key to improving hemodynamic parameter estimation in CKD.
• Further research and standardization are required to facilitate the clinical translation of advanced renal MRI registration techniques.