B0 Correction for 3T Amide Proton Transfer (APT) MRI Using a Simplified Two-Pool Lorentzian Model of Symmetric Water and Asymmetric Solutes

Abstract

Amide proton transfer (APT)-weighted MRI is a promising molecular imaging technique that has been employed in clinic for detection and grading of brain tumors. MTRasym, the quantification method of APT, is easily influenced by B0 inhomogeneity and causes artifacts. Current model-free interpolation methods have enabled moderate B0 correction for middle offsets, but have performed poorly at limbic offsets. To address this shortcoming, we proposed a practical B0 correction approach that is suitable under time-limited sparse acquisition scenarios and for B1 ≥ 1 μT under 3T. In this study, this approach employed a simplified Lorentzian model containing only two pools of symmetric water and asymmetric solutes, to describe the Z-spectral shape with wide and ‘invisible’ CEST peaks. The B0 correction was then performed on the basis of the fitted two-pool Lorentzian lines, instead of using conventional model-free interpolation. The approach was firstly evaluated on densely sampled Z-spectra data by using the spline interpolation of all acquired 16 offsets as the gold standard. When only six offsets were available for B0 correction, our method outperformed conventional methods. In particular, the errors at limbic offsets were significantly reduced (n = 8, p < 0.01). Secondly, our method was assessed on the six-offset APT data of nine brain tumor patients. Our MTRasym (3.5 ppm), using the two-pool model, displayed a similar contrast to the vendor-provided B0-orrected MTRasym (3.5 ppm). While the vendor failed in correcting B0 at 4.3 and 2.7 ppm for a large portion of voxels, our method enabled well differentiation of B0 artifacts from tumors. In conclusion, the proposed approach could alleviate analysis errors caused by B0 inhomogeneity, which is useful for facilitating the comprehensive metabolic analysis of brain tumors.

Keywords: B0 inhomogeneity correction; Lorentzian fitting; amide proton transfer (APT) MRI; brain tumors; chemical exchange saturation transfer (CEST) MRI.

Figure 1

The flowchart of the proposed 2-pool-Lorentzian-based B₀ correction method.

Figure 2

MTR_asym maps and MTR_asym error maps of a representative brain tumor patient. (a) MTR_asym maps without B₀ correction, gold standard, and MTR_asym maps corrected using spline, 1-pool Lorentzian and 2-pool Lorentzian methods; (b) T2w, Gd-T1w and ΔB₀ map; (c) the corresponding MTR_asym error maps with gold standard.

Figure 3

ROI analysis of a representative tumor patient. (a) Four circle ROIs (radius = 5 voxels) with different B₀ inhomogeneity, shown on T2w and ΔB₀ map; (b) mean Z-spectra of ROIs, including gold standard, and Z-spectra corrected using spline, 1-pool Lorentzian, and 2-pool Lorentzian methods; (c) the corresponding Z-spectra error with gold standard.

Figure 4

Statistical analysis of 8 subjects (4 healthy volunteers and 4 brain tumor patients). (a) The statistical results of mean Z-spectra errors corrected using spline, 1-pool Lorentzian and 2-pool Lorentzian methods at 6 offsets. (b) The statistical results of mean MTR_asym error corrected using spline, 1-pool Lorentzian and 2-pool Lorentzian methods; * p < 0.05; ** p < 0.01.

Figure 5

MTR_asym maps and ROI analysis of a meningioma patient. (a) Four circle ROIs (radius = 5 voxels) with different B₀ inhomogeneities, shown on T2w, Gd-T1w, and ΔB₀ map; (b) MTR_asym maps without B₀ correction and MTR_asym maps corrected using vendor, spline, 1-pool Lorentzian and 2-pool Lorentzian methods; (c) mean Z-spectra of ROIs.

Figure 6

A representative brain tumor patient (a,b) and the statistical results of nine brain tumor patients (c). (a) ROIs of high B₀ inhomogeneity regions (overlapped on ΔB₀ map), ROI of tumor (overlapped on Gd-T1w) and the corresponding T2w. (b) MTR_asym maps corrected using the vendor, spline, and 2-pool Lorentzian methods. (c) The statistical results of nine tumor patients, which compared tumor regions with high B₀ inhomogeneity regions on MTR_asym corrected using the vendor, spline and 2-pool Lorentzian methods. Because the vendor filtered many voxels of MTR_asym (4.3 ppm), we excluded MTR_asym (4.3 ppm) provided by the vendor. * p < 0.05; ** p < 0.01.