Artifacts and image quality in diffusion imaging

The artifacts basically stem from the use of:

• diffusion gradients that need to be strong: contraction / dilation and/or shift / distortion of the image (currents induced during gradient ascent and descent), image distortion (errors in gradient linearity), ghost images (gradient instability)
• fast imaging techniques (EPI): magnetic susceptibility effects, ghost images (induced currents).

 

Any gradient-dependent artifact will differ in appearance depending on the b-factor of the image. The ADC map calculation will be perturbed in this case, with measurement errors.

Diffusion-weighted sequences are highly sensitive to macroscopic motions (patient, vascular pulsations…). By virtue of their speed, echo planar sequences will avoid such artifacts. If other, slower types of sequence are used, it may be necessary to synchronize acquisitions with the ECG or with an echonavigator.

 

In diffusion tensor imaging, the main limitation concerns the fiber crossings in the same voxel: the main diffusion direction given by the diffusion tensor does not correspond to a real trajectory and this can lead to errors in a nerve fiber tractography algorithm. The other problem in tractography refers to the fusions, divisions and angulations of the nerve fiber bundles. To overcome these problems, more diffusion measurements can be taken in different directions (HARDI: High Angular Resolution Diffusion Imaging) delivering more data to the algorithms, but at the cost of increased acquisition times.