Toward a realistic in silico abdominal phantom for QSM

Area of Science:

• Medical Imaging
• Biophysics
• Computational Modeling

Background:

• Quantitative Susceptibility Mapping (QSM) is increasingly applied outside the brain, particularly in abdominal imaging.
• Assessing QSM reconstruction algorithms in the abdomen is challenging due to a lack of reliable ground truths and complex signal artifacts (fat, gas, motion).

Purpose of the Study:

• To develop and present a realistic in silico phantom for the creation, evaluation, and comparison of abdominal QSM reconstruction algorithms.
• To provide a standardized tool for benchmarking QSM methods in a challenging anatomical region.

Main Methods:

• Generated synthetic susceptibility and $$ maps from segmented abdominal 3T MRI data of a healthy volunteer.
• Assigned tissue-specific susceptibility and $$ values based on literature and experimental data, incorporating realistic textures and fat contributions.
• Simulated three susceptibility scenarios and two acquisition protocols to test various reconstruction algorithms.

Main Results:

• The developed phantom effectively highlights the strengths and limitations of different QSM acquisition approaches and reconstruction algorithms.
• Identified key areas for improvement in techniques like in-phase acquisitions, water-fat separation, and QSM dipole inversion.

Conclusions:

• The in silico phantom serves as a valuable ground truth for evaluating and comparing abdominal QSM reconstruction pipelines.
• The publicly available, modular source code facilitates customization for diverse abdominal QSM research scenarios.