MRI has proven to be a useful modality to noninvasively diagnose and monitor iron overload disorders.[1] R2* (1/T2*) is commonly estimated using a Cartesian gradient echo scan at 1.5T or 3T. At 1.5T, signal loss limits the dynamic range of liver R2* to liver iron concentrations (LIC) below 35-40 mg Fe/g dry tissue (henceforth, mg/g), which prevents iron quantification in severely loaded patients. At 3T, signal loss is twice as rapid, preventing LIC estimation in half of the clinically relevant range. UTE (ultrashort echo time) imaging shows promise as a method to image high-iron patients who demonstrate rapid T2* decay in the liver. We present data demonstrating that a readily implementable UTE sequence can extend the dynamic range of R2* quantitation at 3T to LIC values exceeding 40 mg/g in patients with iron overload.

A UTE radial protocol was performed on a Philips Achieva 3T magnet using the 16-element SENSE Torso XL coil (R3.2.2, Philips Healthcare, Best, Netherlands). Six transverse slices were acquired using a slab-selective UTE excitation followed by a 3D stack-of-stars radial readout with the following parameters: matrix=88x88, voxel=3.5x3.5x15mm, TR=4ms, NEX=1, Flip angle=5°, TFE Factor=200, angle density=160%, TE=[0.19,0.23,0.35,0.60, 0.85,1.0,2.0] ms. In human subjects, clinical assessment was performed on a 1.5T Achieva (SW R3.2.2, Philips) with a SENSE Torso XL coil. A 3-slice single gradient echo acquisition was performed with the following parameters: 16 echoes linearly spaced echoes with TE=0.96-11.47ms, FA=30°, matrix=72x67, voxel=1.25x1.25x10mm, BW=4409Hz/px.

The UTE sequence was tested in both phantoms and iron-overloaded study participants. The phantom contained vials of [0,0.50,0.75,1.0,1.5,2.0,2.5,3.5,5.0,8.0,12,16,24] mM manganese chloride in a 0.25 mM MnCl2 bath to simulate a variety of rapid transverse decay rates. Results were compared to known transverse decay rates for each concentration. Five human subjects (3M/2F, 7-23 years, 4 β-Thalassemia, 1 other anemia) underwent clinically indicated MR assessment at 1.5T as well as the free-breathing UTE 3T research protocol as part of an IRB-approved, NIH-funded study. To facilitate comparison of R2* values, all fitting of data was performed on magnitude images using a mono-exponential signal model with manual echo truncation with software developed in MATLAB (Mathworks, Natick, MA).

Relaxation rate by UTE was linear with manganese chloride concentration over the entire dynamic range (r2=0.996, Figure 1). The observed slope of 116.8 Hz/mM is very close to 1.41 times the empirically determined rate multiplier of 78 Hz/mM, which we have previously measured at 1.5T. These results closely matched previous unreported R2* estimates in this phantom at 3T.

Patient results (Figure 2a) also demonstrated linear correlation when compared to a clinical R2* estimate with an R-squared of 0.9853. The slope nearly matches the theoretical multiplier of 2 that is expected between 1.5T and 3T R2* estimates with iron-mediated dephasing.[2] The Bland-Altman analysis (Figure 2b) shows no significant bias between the two methods over a clinical range of 2-38 mg/g.

Using UTE imaging, we have overcome previous limitations on the dynamic range of iron imaging at 3T. This method will quantify the entire clinical range of liver iron overload. The patient data demonstrated a near doubling of R2* compared to 1.5T estimates. This relationship was previously shown using a Monte Carlo simulation framework and patient data.[2]

R2* is fundamentally a better contrast for iron quantification than R2 because R2*-based LIC estimates are not influenced by B1+ inhomogeneity errors that cause iron overestimates [3] and are more robust to changes in tissue iron concentration.[4] This UTE method extends 3T R2* linearity with iron to all achievable iron burdens using a pulse sequence currently available on existing platforms.

Slab-selective UTE imaging with a 3D SOS radial acquisition shows promise as a technique for clinical R2* estimation at 3T. The sequence is expected to reach R2* values of over 4,000 s-1, or over 60 mg/g. Validation of this technique will continue as part of a patient study to improve iron overload imaging at 3T. We also hope to leverage complex data fitting to reduce noise bias.