Limits...
A method for improving the performance of gradient systems for diffusion-weighted MRI.

Nagy Z, Weiskopf N, Alexander DC, Deichmann R - Magn Reson Med (2007)

Bottom Line: This effect can be increased by the use of large, balanced bipolar gradients.It was found that the gradient axes (+/-x, +/-y, +/-z) were calibrated differently, resulting in offset ADC values.In addition, fiber tracking results in the human brain were noticeably affected by improving the gradient system calibration.

View Article: PubMed Central - PubMed

Affiliation: Wellcome Trust Centre for Neuroimaging, Institute of Neurology, University College London, London, UK. z.nagy@fil.ion.ucl.ac.uk

ABSTRACT
The MR signal is sensitive to diffusion. This effect can be increased by the use of large, balanced bipolar gradients. The gradient systems of MR scanners are calibrated at installation and during regular servicing visits. Because the measured apparent diffusion constant (ADC) depends on the square of the amplitude of the diffusion sensitizing gradients, errors in the gradient calibration are exaggerated. If the error is varying among the different gradient axes, it will affect the estimated degree of anisotropy. To assess the gradient calibration accuracy in a whole-body MRI scanner, ADC values were calculated for a uniform water phantom along each gradient direction while monitoring the temperature. Knowledge of the temperature allows the expected diffusion constant of water to be calculated independent of the MRI measurement. It was found that the gradient axes (+/-x, +/-y, +/-z) were calibrated differently, resulting in offset ADC values. A method is presented to rescale the amplitude of each of the six principal gradient axes within the MR pulse sequence. The scaling factor is the square root of the ratio of the expected and observed diffusion constants. In addition, fiber tracking results in the human brain were noticeably affected by improving the gradient system calibration.

Show MeSH
ADC values along different gradient directions. Two different experiments are described each performed without (top) and with (bottom) gradient corrections. The y-axis is scaled identically for all plots in units of mm2/s. In (a) and (c) the ADC values were calculated from seven images with b = 100 s/mm2 and 61 noncollinear directions distributed on the surface of a hemisphere with b = 1000 s/mm2 (for uniformity only 60 are shown). In (b) and (d) the ADC values were calculated from seven images with b = 100 s/mm2 and 10 images along both the positive and negative direction of each of the physical gradient axes with b = 1000 s/mm2 (see the top of each plot for indication of the gradient). The experiments were performed on a water phantom where isotropic diffusion is expected. a: Shows a high degree of variability in ADC values along the different directions. The variance is much higher than would be expected from the SNR of the images. b: Demonstrates that the variability in (a) is due in a large part to a systematic difference of ADC values along the different gradient axes. c and d: The results of the same two experiments as in (a) and (b) respectively after the gradient amplitudes were rescaled based on the methods described in this work (using Eq. [6]).
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2683063&req=5

fig01: ADC values along different gradient directions. Two different experiments are described each performed without (top) and with (bottom) gradient corrections. The y-axis is scaled identically for all plots in units of mm2/s. In (a) and (c) the ADC values were calculated from seven images with b = 100 s/mm2 and 61 noncollinear directions distributed on the surface of a hemisphere with b = 1000 s/mm2 (for uniformity only 60 are shown). In (b) and (d) the ADC values were calculated from seven images with b = 100 s/mm2 and 10 images along both the positive and negative direction of each of the physical gradient axes with b = 1000 s/mm2 (see the top of each plot for indication of the gradient). The experiments were performed on a water phantom where isotropic diffusion is expected. a: Shows a high degree of variability in ADC values along the different directions. The variance is much higher than would be expected from the SNR of the images. b: Demonstrates that the variability in (a) is due in a large part to a systematic difference of ADC values along the different gradient axes. c and d: The results of the same two experiments as in (a) and (b) respectively after the gradient amplitudes were rescaled based on the methods described in this work (using Eq. [6]).

Mentions: For experiment 1, ADC values along 60 different directions are shown in Fig. 1a. Contrary to the expectation of spatially uniform ADC values, there is a high degree of variability in this measurement. The percent difference between the minimum and maximum ADC values from this measurement is 6.9%. In contrast the mean coefficient of variation of the 60 regions of interest was 4.3%, leading to a standard error of the mean of about 0.4%.Figure 1


A method for improving the performance of gradient systems for diffusion-weighted MRI.

Nagy Z, Weiskopf N, Alexander DC, Deichmann R - Magn Reson Med (2007)

ADC values along different gradient directions. Two different experiments are described each performed without (top) and with (bottom) gradient corrections. The y-axis is scaled identically for all plots in units of mm2/s. In (a) and (c) the ADC values were calculated from seven images with b = 100 s/mm2 and 61 noncollinear directions distributed on the surface of a hemisphere with b = 1000 s/mm2 (for uniformity only 60 are shown). In (b) and (d) the ADC values were calculated from seven images with b = 100 s/mm2 and 10 images along both the positive and negative direction of each of the physical gradient axes with b = 1000 s/mm2 (see the top of each plot for indication of the gradient). The experiments were performed on a water phantom where isotropic diffusion is expected. a: Shows a high degree of variability in ADC values along the different directions. The variance is much higher than would be expected from the SNR of the images. b: Demonstrates that the variability in (a) is due in a large part to a systematic difference of ADC values along the different gradient axes. c and d: The results of the same two experiments as in (a) and (b) respectively after the gradient amplitudes were rescaled based on the methods described in this work (using Eq. [6]).
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2683063&req=5

fig01: ADC values along different gradient directions. Two different experiments are described each performed without (top) and with (bottom) gradient corrections. The y-axis is scaled identically for all plots in units of mm2/s. In (a) and (c) the ADC values were calculated from seven images with b = 100 s/mm2 and 61 noncollinear directions distributed on the surface of a hemisphere with b = 1000 s/mm2 (for uniformity only 60 are shown). In (b) and (d) the ADC values were calculated from seven images with b = 100 s/mm2 and 10 images along both the positive and negative direction of each of the physical gradient axes with b = 1000 s/mm2 (see the top of each plot for indication of the gradient). The experiments were performed on a water phantom where isotropic diffusion is expected. a: Shows a high degree of variability in ADC values along the different directions. The variance is much higher than would be expected from the SNR of the images. b: Demonstrates that the variability in (a) is due in a large part to a systematic difference of ADC values along the different gradient axes. c and d: The results of the same two experiments as in (a) and (b) respectively after the gradient amplitudes were rescaled based on the methods described in this work (using Eq. [6]).
Mentions: For experiment 1, ADC values along 60 different directions are shown in Fig. 1a. Contrary to the expectation of spatially uniform ADC values, there is a high degree of variability in this measurement. The percent difference between the minimum and maximum ADC values from this measurement is 6.9%. In contrast the mean coefficient of variation of the 60 regions of interest was 4.3%, leading to a standard error of the mean of about 0.4%.Figure 1

Bottom Line: This effect can be increased by the use of large, balanced bipolar gradients.It was found that the gradient axes (+/-x, +/-y, +/-z) were calibrated differently, resulting in offset ADC values.In addition, fiber tracking results in the human brain were noticeably affected by improving the gradient system calibration.

View Article: PubMed Central - PubMed

Affiliation: Wellcome Trust Centre for Neuroimaging, Institute of Neurology, University College London, London, UK. z.nagy@fil.ion.ucl.ac.uk

ABSTRACT
The MR signal is sensitive to diffusion. This effect can be increased by the use of large, balanced bipolar gradients. The gradient systems of MR scanners are calibrated at installation and during regular servicing visits. Because the measured apparent diffusion constant (ADC) depends on the square of the amplitude of the diffusion sensitizing gradients, errors in the gradient calibration are exaggerated. If the error is varying among the different gradient axes, it will affect the estimated degree of anisotropy. To assess the gradient calibration accuracy in a whole-body MRI scanner, ADC values were calculated for a uniform water phantom along each gradient direction while monitoring the temperature. Knowledge of the temperature allows the expected diffusion constant of water to be calculated independent of the MRI measurement. It was found that the gradient axes (+/-x, +/-y, +/-z) were calibrated differently, resulting in offset ADC values. A method is presented to rescale the amplitude of each of the six principal gradient axes within the MR pulse sequence. The scaling factor is the square root of the ratio of the expected and observed diffusion constants. In addition, fiber tracking results in the human brain were noticeably affected by improving the gradient system calibration.

Show MeSH