MR Toolkit - Recon and Parallel Imaging

In the rightmost image above, how is the brain oriented?

• A. The image shows a transverse/axial slice of a brain.
• B. The image shows a coronal slice of a brain.
• C. The image shows a sagittal slice of a brain.

Look at the Image with reduced FOV and the Image by cropping edges. Can you explain why the Image with reduced FOV looks noisier? Think about it for a minute before answering the question below.

The reduction of the FOV can be done…

• A. By subsampling the k-space. In the example image, a combined reduction factor of 24 is possible.
• B. By sampling a central region of k-space. A total factor of 4 could be reduced without sacrificing on the resolution.
• C. Either in k-space or directly in the image space by simply cropping the image. The results are equivalent. In the example image a reduction factor of 24 could be achieved.

To increase the SNR available in the dataset with reduced FOV, it is better to have a lower resolution than currently prescribed. The original resolution was 0.85mm. Let us make it 1.7mm on both x and y.

How does the new low resolution image look like?

• A. The image looks just as noisy as before.
• B. The image is less noisy than before but some details are now unclear.
• C. Because the image already had a tight FOV it wasn't possible to increase the SNR.

Formulate the dependence of SNR as a function of: Number of phase encoding steps,\( N_{PE}\) and voxel volume, \(V\);

• A. \(\text{SNR}_{} = {Signal(TR,TE,\alpha) \bullet V \bullet N}_{\text{PE}}\)
• B. \(\text{SNR}_{} = Signal(TR,TE,\alpha) \bullet \sqrt{V \bullet N_{\text{PE}}}\)
• C. \(\text{SNR}_{} = Signal(TR,TE,\alpha) \bullet V\sqrt{N_{\text{PE}}}\)

How is information distributed in k-space? Consider the k-space representation of the image of reduced FOV and resolution of the previous exercise. What information do you think was present on the k-space coordinate (kx=ky=10)?

• A
• B
• C
• D
• E

Parallel imaging can be used to accelerate the acquisition of single slice. In the case of structural imaging it shortens the acquisition time. In case of echo planar imaging, EPI, it is used to reduce the echo train length and distortions. The images will suffer an SNR penalty because of the reduced number of phase encoding steps measured (see the answer to question 3), and also a so-called G-factor penalty because of noise amplification associated with disentangling overlapping pixels. The SNR of an accelerated acquisition with an acceleration factor of AF can thus be written as:

\(\text{SN}R_{\text{af}} = \frac{\text{SNR}}{\sqrt{\text{af}} \times g_{\text{factor}}}\)

Look at the following examples:

Which acceleration factor was used to acquire the image below?

• A. Acceleration factor 2.
• B. Acceleration factor 3.
• C. Acceleration factor 4.
• D. Acceleration factor 5.

Consider the case where three slices were simultaneously excited. The acquisition was done with six receiver coils for which the following sensitivity maps had been acquired:

When encoding the acquisition, three different CAIPI factors were applied:

1 - no FOV shift;	2 - 0.5xFOV shift;	3 - 0.33xFOV shift

Which of the data sets would be better reconstructed using standard parallel imaging?

• A. The three data sets can be equally well reconstructed because the same information is present in all of them.
• B. Because the six coil sensitivities used don't have much variations across the slice direction, the reconstruction will fail independent of shifts.
• C. The reconstruction for the no FOV shift will fail, and the remaining ones will be better.

The in-plane acceleration (along the phase encoding direction) achievable without significant g-factor penalty is independent of the usage of simultaneous multi-slice excitation.

• A. True
• B. False

---