Potential in reducing scan times of HARDI by accurate correction of the cross‐term in a hemispherical encoding scheme

Abstract

Purpose: To reduce the scan time of high angular resolution diffusion imaging (HARDI) by using the hemispherical encoding scheme with the cross-term correction. Materials and Methods:Unidirectional and 45°crossing phantoms were built to evaluate the accuracy of the fiber orientation estimation when using a hemispherical encoding scheme with and without the cross-term correction. The q-ball imaging using the spherical harmonic basis was adopted for estimation of fiber orientation. By recalculating the diffusion gradient strengths, we make the actual b value in all directions equal to the expected b value for the correction of the cross-term. Results:The separation angles measured on the crossing phantom are 42.67°Ϯ 1.50°, 42.75°Ϯ 2.10°, and 42.93°Ϯ 2.78°for cross-term-free results, for the results from the hemispherical encoding scheme with and without the crossterm correction. The angular errors in mapping the unidirectional phantom are 1.87°-2.29°and 4.01°-4.92°for the results using the hemispherical encoding scheme with and without correcting for the cross-term at different b values. Conclusion:Using the hemispherical encoding scheme with the cross-term correction can potentially halve the scan time of HARDI and obtain accurate fiber orientation estimation simultaneously. This will be helpful in the clinical application of HARDI. THE DEVELOPMENT OF diffusion nuclear magnetic resonance (NMR) enables us to noninvasively retrieve microstructural information that can hardly be detected using conventional NMR methods (1,2). Diffusion tensor imaging (DTI) was later proposed to measure the fiber orientation by estimating threedimensional probability distribution using a Gaussian approximation (3). This opened the door to advances in white matter (WM) connectivity (4 -8). Even though DTI can well assess the integrity of axonal fibers and accurately define the fiber orientation within a voxel containing fibers with a coherent orientation, its drawback is that the fiber orientation estimate is ambiguous in voxels containing multiple orientation structures. This drawback was highlighted in human cerebral studies with the millimeter width of a typical magnetic resonance imaging (MRI) voxel (9,10). In order to address the inability of DTI to resolve intravoxel fiber crossings, several reconstruction methods based on the high angular resolution diffusion imaging (HARDI) sampling scheme (11-17) were proposed and assessed to resolve intravoxel fiber crossings via numerical phantoms and in vivo studies. However, long acquisition times due to the need to obtain a great deal of diffusion-weighted images (DWIs) in HARDI (Ϸ60 -400 DWIs) limit the possibility of routine HARDI and its clinical applications.In previous studies, shortening of the acquisition times of HARDI was achieved using an improved reconstruction method with a lower number of DWIs. Khachaturian et al (18) boosted a sampling efficiency gain of 274%-377% for q-ball imaging (QBI) by nonlinearly fusing the diffusion signal from separate low and high wavevector acquis...

The effect of finite diffusion gradient pulse duration on fibre orientation estimation in diffusion MRI

“…the numerical, the water capillary, and the sciatic nerve phantom data) are displayed using The crossing fibre data were reconstructed using QBI (Tuch, 2004) to provide the orientation distribution function (ODF), with angular sampling resolution of 4°(16-fold tessellated icosahedrons). To reduce the error caused by the discrete spherical sampling of ODF, we randomly rotated the angular sampling scheme and computed the ODF 1000 times for each voxel (Cho et al, 2009). During each ODF calculation, we considered local maximum vectors of each ODF with lengths greater than 50% of the peak ODF value as the potential fibre directions.…”

Development, integration and use of an ultra-high-strength gradient system on a human-size 3 T magnet for small animal MRI

“…Although the diffusion cross-term is eliminated by cross-term correction using (3), the constant term due to the imaging gradient cannot be removed. To minimize the error in mapping the ADC value, the constant term from the imaging gradient must be considered in the calculation of the b-value [13, 14], as shown in Table 1. The cross-term-free diffusion signal and the exact b-values were then applied to a mono-exponential decay model for ADC calculation: , where S is the diffusion-weighted signal, and S 0 is the non-diffusion-weighting signal.…”