Evaluation of the accuracy and angular resolution of q-ball imaging

TY - JOUR

T1 - Evaluation of the accuracy and angular resolution of q-ball imaging

AU - Cho, Kuan-Hung

AU - Yeh, Chun-Hung

AU - Tournier, Jacques-Donald

AU - Chao, Yi-Ping

AU - Chen, Jyh-Horng

AU - Lin, Ching-Po

PY - 2008/8/1

Y1 - 2008/8/1

N2 - Q-ball imaging (QBI) has been proposed for the mapping of multiple intravoxel fiber structures using the Funk-Radon transform on high angular resolution diffusion images (HARDI). However, the accuracy and the angular resolution of QBI to define fiber orientations and its dependence on diffusion imaging parameters remain unclear. The phantom models, made up of sheets of parallel capillaries filled with water, were designed to evaluate the accuracy and the angular resolution of QBI at different vertical bar q vertical bar values. With an inner diameter of 20 mu m and an outer diameter of 90 mu m, the capillaries afforded a restrictive environment compared with the diffusion measurement scale. Further, the angular resolutions of QBI at various vertical bar q vertical bar value were also quantified on the corpus callosum in the human brain. The full width at half maximum (FWHM) of the main lobe of normalized orientation distribution function (nODF) was calculated and adopted to quantify the angular resolution of QBI. With the phantom model, a higher vertical bar q vertical bar value resulted in worse accuracy but better angular resolution for QBI. The same trend where a higher vertical bar q vertical bar value yielded a better angular resolution was also observed in the human study. Upon comparison of QBI with T2W1, QBI with vertical bar q vertical bar=277 cm(-1) (b=3000 s/mm(2)) was found to be insufficient to differentiate capillaries crossing at 45 degrees. However, when encoding with vertical bar q vertical bar=320, 358, and 392 cm(-1) (b=4000, 5000, and 6000 s/mm(2)), the deviation angles between the primary ODF and the 45 degrees phantoms were -4.91 degrees +/- 2.72 degrees, -1.37 degrees +/- 2.32 degrees, and -0.69 degrees +/- 1.54 degrees with adequate signal-to-noise ratio (SNR). These results were consistent with the FWHM-nODF, which showed that a vertical bar q vertical bar value of 320 cm(-1) was the threshold to resolve capillaries intersecting at 45 degrees. Additionally, it was demonstrated in both the phantom model and the human brain that QBI encoding with lower vertical bar q vertical bar values may result in underestimation of the orientations of the crossing fibers. In conclusion, QBI was found to accurately resolve crossing fiber orientations and was highly dependent on the selected vertical bar q vertical bar value.

AB - Q-ball imaging (QBI) has been proposed for the mapping of multiple intravoxel fiber structures using the Funk-Radon transform on high angular resolution diffusion images (HARDI). However, the accuracy and the angular resolution of QBI to define fiber orientations and its dependence on diffusion imaging parameters remain unclear. The phantom models, made up of sheets of parallel capillaries filled with water, were designed to evaluate the accuracy and the angular resolution of QBI at different vertical bar q vertical bar values. With an inner diameter of 20 mu m and an outer diameter of 90 mu m, the capillaries afforded a restrictive environment compared with the diffusion measurement scale. Further, the angular resolutions of QBI at various vertical bar q vertical bar value were also quantified on the corpus callosum in the human brain. The full width at half maximum (FWHM) of the main lobe of normalized orientation distribution function (nODF) was calculated and adopted to quantify the angular resolution of QBI. With the phantom model, a higher vertical bar q vertical bar value resulted in worse accuracy but better angular resolution for QBI. The same trend where a higher vertical bar q vertical bar value yielded a better angular resolution was also observed in the human study. Upon comparison of QBI with T2W1, QBI with vertical bar q vertical bar=277 cm(-1) (b=3000 s/mm(2)) was found to be insufficient to differentiate capillaries crossing at 45 degrees. However, when encoding with vertical bar q vertical bar=320, 358, and 392 cm(-1) (b=4000, 5000, and 6000 s/mm(2)), the deviation angles between the primary ODF and the 45 degrees phantoms were -4.91 degrees +/- 2.72 degrees, -1.37 degrees +/- 2.32 degrees, and -0.69 degrees +/- 1.54 degrees with adequate signal-to-noise ratio (SNR). These results were consistent with the FWHM-nODF, which showed that a vertical bar q vertical bar value of 320 cm(-1) was the threshold to resolve capillaries intersecting at 45 degrees. Additionally, it was demonstrated in both the phantom model and the human brain that QBI encoding with lower vertical bar q vertical bar values may result in underestimation of the orientations of the crossing fibers. In conclusion, QBI was found to accurately resolve crossing fiber orientations and was highly dependent on the selected vertical bar q vertical bar value.

KW - q-ball

KW - phantom model

KW - optimization

KW - DIFFUSION-WEIGHTED MRI

KW - MAGNETIC-RESONANCE

KW - HUMAN BRAIN

KW - TENSOR MRI

KW - FIBER

KW - VALIDATION

KW - TRACKING

KW - SPECTROSCOPY

KW - COEFFICIENTS

KW - ANISOTROPY

U2 - 10.1016/j.neuroimage.2008.03.053

DO - 10.1016/j.neuroimage.2008.03.053

M3 - Article

SN - 1053-8119

VL - 42

SP - 262

EP - 271

JO - NeuroImage

JF - NeuroImage

IS - 1

ER -