DIFFUSION WEIGHTED IMAGING
Spiral based diffusion weighted imaging
Variable Density Spiral (VDS) based DTI
Parallel imaging and compressed sensing combined framework for accelerating high-resolution diffusion tensor imaging using inter-image correlation. parallel imaging (PI) and compressed sensing (CS) combined framework is proposed, which features motion error correction, PI calibration, and sparsity model using inter-image correlation tailored for high-resolution DTI, named anisotropic sparsity SPIRiT (AS-SPIRiT). A specific implementation based on multishot variable density spiral DTI is used to demonstrate the method. Compared with CG-SENSE, CS-based joint reconstruction, and PI and CS combined methods with L1 and joint sparsity regularization, AS-SPIRiT resulted in better preserved image quality and more accurate DTI parameters than other methods (R=3 in the figure below).
Navigator-Free Regular Spiral DTI
A new proposed reconstruction method, named POCS-enhanced Inherent Correction of motion-induced phase Errors (POCS-ICE), can reliably reconstruct multi-shot spiral DWI images without use of navigator, for even relatively high shot number. In vivo experiments show the improved image quality in terms of higher SNR and less artifacts compared with SENSE+CG. The improved spatial resolution and image resolvability are beneficial to study brain micro-structures and physiological features for neuroscience.
Common-Information enhanced high resolution VDS DTI reconstruction
Multi-shot VDS is capable of achieving high resolution DTI, but requires long scan time. Conventional fast VDS DTI method, CG-SENSE is limited by SNR penalty. Common information shared by different directions is used to optimize sampling trajectory and enhance image reconstruction. A rotated interleaved acquisition scheme is proposed to provide extra complementary information. SPIRiT is used as a specific reconstruction example in our implementation.
High Resolution Diffusion Weighted Imaging Using interleaved EPI Sequence
Multi-shot Interleaved EPI (iEPI) sequence has faster k-space traversal speed than ssEPI, so less distortion and T2* blurring happens. iEPI can be used for high resolution DWI or DTI. For multishot acquisitions, diffusion sensitizing gradient is sensitive to motion and will introduce different phase in different excitation. Direct reconstruction will introduce ghost artifacts.
Our self-feeding MUSE with data rejection can achieve artifact-free high resolution diffusion weighted images without extra navigators.
Black Blood Cine Imaging of Common Carotid Artery
Artery dynamics are clinically significant for indicating cardiovascular disease since it can reveal functional characteristics of vasculature. Bright blood cine imaging has been used to measure common carotid artery (CCA) dynamics, but is susceptible to flowing artifacts.
Black blood imaging can reduce flowing artifacts by suppressing blood signals, so it is believed that black blood cine (BB-CINE) imaging can provide sharper border definition and more accurate quantification for arterial dynamics. In our study, MSDE prepared BB-CINE imaging technique is proposed and implemented. This method has been proved to provide better contrasts between blood and vessel wall. Thus this method is believed to be more promising for investigating functional properties of the vasculature, monitoring vessel stiffness changes, and predicting related cardiovascular diseases.
Figure. 1. Zoomed BB-CINE and bright blood images of left CCA at different cardiac phases (15 phases in total). Bright blood images may suffer from flowing artifacts (red arrows), which were absent for BB-CINE images.
Comparison of Different Reference Selection Strategies for PROPELLER MRI
Three different reference selection methods were implemented and compared.
SBR: Single Blade Reference, use a best blade as the reference.
CBR: Combined Blade Reference, take the weighted average of all blades as the reference.
GBR: Grouped Blade Reference, blades are classified into groups, and the blades in the dominant group are averaged to generate the reference.
Fig. 1. Illustration of SBR, CBR and GBR.
The motion estimation and correction process and parameters were carefully tuned, and no iteration was performed for all selection strategies above. Images of similar quality can be generated using all the strategies above, with motion artifacts significantly eliminated.
Fig. 2. In vivo results using different reference selection strategies.