Investigation on the optimization approaches of diffusion weighted imaging

The corticospinal tract is important in the guidance of neurosurgery. Therefore precise tractography in the pre-operative plan is necessary. However, the inherent drawback of DWI in image acquisition makes it easy to be affected by bulk motion and pulsatile motion and also to produce image distortions because of EPI acquisitions. Therefore, optimized approaches aimed at reducing or eliminating these artifacts and improve image quality have been investigated. Pulsatile motion occurs during the cardiac systolic period and has been reported to produce motion artifacts in the brain stem and basal ganglia, which might affect the corticospinal tract. Up to now, there is no consensus on the real effect of pulsatile motion on the diffusion properties, diffusion tensor parameters and fiber tractography, and the role of cardiac gating to overcome these effects is also not very clear. So in part 1 of the current study, we analyzed the influence of pulsatile motion and the contribution of cardiac-gating in the improvement of the quality of DWI, DTI and tractography. We found obvious signal attenuation in the brain stem and cerebellum. Pulsatile motion led to an over-estimation of FA and under-estimation of MD along the CST. Cardiac-gating could help to reduce the bias of the diffusion tensor parameters. Although pulsatile motion resulted in motion artifacts, bias of the diffusion tensor parameters and deviation of the principal eigenvector direction, it did not influence tract volume and location when a deterministic algorithm was applied for the reconstruction of the tract. Therefore, in this part we knew that cardiac-gating could help to avoid the motion artifacts and bias of the diffusion tensor parameters. But for the tractography of CST, the current image acquisition methods with high angular resolution or averaging seemed already able to overcome the effects of pulsatile motion, and cardiac-gating can’t make significant contribution. In part 2 of this study, we focused on another approach for improving the DWI image quality, the denoising algorithm POAS (Position-orientation adaptive smoothing). The DWI suffers more easily from artifacts during acquisition and always has a low SNR, which might lead to erroneous decisions in the determination of the diffusion metrics and fiber tractography in clinics. Although plenty of denoising methods have been proposed up to now, POAS came into consideration because POAS reduces image noise in the whole brain with edge-preserving properties and avoids blurring. In this study, we found that POAS reduced noise directly on DWIs and improved SNR dramatically, and consequently, POAS also reduced the bias and variation of the diffusion tensor quantities, such as FA. In tractography, after processing with POAS, a greater fiber volume of the CST was reconstructed compared to the original datasets. At the same time, reconstruction of the CST in POAS-processed datasets gained more stability and less variability which could compensate for the effect of a high angular resolution in some degree. In the future, the application of POAS in pathological cases should be conducted to verify its practical value in the clinics. In neuroscience, the image quality of DWI and the precision of the diffusion tensor parameters are essential. Both of the above approaches could be applied to optimize the analysis. During neurosurgical operations, the accuracy of tract reconstruction, or space occupation, has more importance. So POAS could be considered to improve tractography while cardiac-gating did not have significant effects. More advanced approaches should be further investigated