Inherent eddy current compensation for high-resolution DTI using bipolar diffusion gradients

Introduction: Clinical diffusion-weighted imaging (DWI) often employs a single-shot echo-planar imaging (EPI) readout to increase acquisition speed. This comes at a cost, however, since conventional DWI sequences employ a spin echo (SE) sequence with either unipolar or bipolar diffusion-sensitizing gradients on both sides of the 180º refocusing pulse. This configuration shortens the separation between strong diffusion gradient pulses and the EPI readout and forces partial k-space acquisition. The close proximity of strong diffusion gradients and the EPI readout increases direction dependent eddy current artifacts. while the reduction in k-space coverage decreases SNR. Here we propose a novel single-sided bipolar (SSBP) SE DWI sequence that reduces this eddy current artifact and improves SNR by separating diffusion-sensitizing gradients from the EPI readout and acquiring full k-space EPI data with reduced TE as compared to a conventional twice-refocused spin echo (TRSE) diffusion preparation DWI technique [1]. Materials and Methods: Healthy volunteers were scanned on a commercial GE 3T MR750 scanner using the newly developed SSBP-DWI and the TRSE-DWI sequences with the same scan parameters: 20 slices with 5mm slice thickness, a 220x220 mm 2 FOV with 192x192 imaging matrix, TE=92ms, TR=10s, parallel imaging (2x acceleration), b=800 mm 2 /s, 20 directions and echo spacing of 944s. The echo train length was 96 for SSBP-DWI and 64 for TRSE-DWI. Results: Our preliminary results indicate that, using SSBP diffusion preparation, it is possible to obtain high-resolution images with superior SNR and significantly reduced eddy current distortions compared to the TRSE-DWI. In the SSBP-DWI baseline (non-diffusion weighted) images, the SNR was 8.7% higher than in the corresponding TRSE-DWI image, due to the use of full k-space acquisition in SSBP-DWI. More importantly, SSBP-DWIs showed virtually no eddy current distortion while the TRSE-DWIs exhibited prominent residual eddy current distortion at the periphery of the brain (Fig. 2), most likely due to the imperfect eddy current compensation, which assumes a mono-exponential decay. In addition to reducing eddy current distortion due to the bipolar nature of the diffusion-sensitizing gradients[2], SSBP-DWI separates strong diffusion gradients and the EPI readout without increasing TE. It is important to note that for large b-values, SSBP-DWI requires longer TEs as compared to TRSE-DWI, potentially resulting in lower SNR. However, this issue can be resolved with availability of stronger gradients. Discussion: To minimize off-resonance distortions from static B 0 field inhomogeneities, the full k-space EPI readout duration (and implicitly TE) can be further reduced e.g., by …