Modeling of T2* decay in vertebral bone marrow fat quantification

Area of Science:

• Magnetic Resonance Imaging
• Biomedical Engineering
• Radiology

Background:

• Bone marrow fat fraction mapping is crucial for assessing bone health and cancer treatment effects.
• Vertebral bone marrow's short T2* relaxation times can interfere with accurate fat quantification.
• Chemical shift encoding-based water-fat separation is a key imaging technique.

Purpose of the Study:

• To compare different T2* correction methods for chemical shift encoding-based water-fat imaging in vertebral bone marrow.
• To evaluate these methods against single-voxel magnetic resonance spectroscopy (MRS) as a reference standard.
• To identify the optimal T2* correction approach for reliable fat fraction mapping.

Main Methods:

• Eight-echo gradient-echo imaging and single-voxel MRS were performed on the L3-L5 spine of 25 healthy volunteers.
• Four T2* correction approaches were evaluated: single-T2*, dual-T2*, T2' with known T2 from MRS, and T2' with average known T2.
• Quantitative analysis involved linear regression comparing imaging-derived and MRS-derived fat fractions.

Main Results:

• Dual-T2* correction yielded noisier fat fraction maps compared to single-T2* or T2' correction.
• Single-T2* and dual-T2* corrections resulted in regression slopes significantly different from 1.
• T2' correction using a priori-known T2 demonstrated a slope not significantly different from 1, with minimal intercept bias and high R(2) values.

Conclusions:

• T2' correction utilizing a priori-known T2 effectively removes fat fraction bias caused by T2* differences between water and fat.
• This method maintains excellent noise performance in vertebral bone marrow fat fraction mapping.
• Accurate fat quantification via T2' correction enhances the utility of MRI in bone health and oncology research.