Optimised passive marker device visibility and automatic marker detection for 3-T MRI-guided endovascular...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Interventional Radiology

Background:

• Passive paramagnetic markers on MRI-compatible endovascular devices create susceptibility artifacts for MRI visibility and guidance.
• Optimizing marker visibility is crucial for automatic detection and real-time tracking of devices during procedures.
• Visibility depends on MRI technical parameters and marker characteristics, necessitating thorough evaluation.

Purpose of the Study:

• To assess marker visibility and automatic detection robustness under varying MRI parameters and marker characteristics.
• To evaluate the impact of gradient-echo (GRE) and balanced steady-state free precession (bSSFP) sequences on marker artifact characteristics.
• To investigate the performance of a deep learning model for automatic marker detection with different imaging parameters.

Main Methods:

• Guidewires with varying iron(II,III) oxide nanoparticle (IONP) concentrations were imaged using GRE and bSSFP sequences at 3 T.
• Parameters such as echo time (TE), slice thickness (ST), and phase encoding direction (PED) were systematically varied.
• Artifact width, contrast-to-noise ratios, image quality scores, and deep learning detection performance were evaluated.

Main Results:

• Artifact width was significantly larger with bSSFP compared to GRE sequences and increased with TE and IONP concentration.
• GRE sequences yielded higher image quality scores than bSSFP.
• While a deep learning model achieved automatic marker detection, its performance decreased with altered PED, TE, and IONP concentration.

Conclusions:

• Adjusting TE and IONP concentration effectively modifies artifact size for sufficient marker visibility.
• Deep learning-based automatic marker detection is feasible but sensitive to changes in MR parameters.
• Consideration of these factors is essential for optimizing device visibility and ensuring reliable automatic detection in MRI-guided endovascular interventions.