NetForum uses cookies to ensure that we give you the best experience on our website. If you continue to use the site, we'll assume that you are happy to receive these cookies on the NetForum website. Read about our cookies.
NetForum Community
Learn. Share. Optimize.
Log in | Sign up now | Submit content | Contact
Go to similar content

Tips for body diffusion weighted imaging (DWI)

Application Tip
Worthington, Paul, BS, RT(R)(MR) MRSO Philips Healthcare • USA

Diffusion weighted imaging (DWI) in body applications

This is an update of the original 2010 Application Tip on Body diffusion weighted imaging. It provides tips on DWI for users of Release 5 and later software.

Diffusion weighted imaging (DWI) is typically a single shot echo-planar based sequence that demonstrates pathology based on fluid motion states at the cellular level. Using a single spin-echo RF excitation, k-space is filled by rapid positive/negative oscillation of the readout gradient. Diffusion gradient pre-pulses are applied based on the desired b-values to sensitize the image to abnormal fluid motion states created by pathology.

Unlike brain imaging where b-value (b = 1000 s/mm2) is rarely manipulated, body protocols commonly require adjustments to b-value based on the clinical requirements of the scan.

EPI-based acquisitions are well suited to high temporal resolution required for imaging of water motion at the cellular level. However, EPI is sensitive to magnetic susceptibility artifacts created by air/soft tissue interfaces in the torso. Chemical shift artifacts are also magnified due to the EPI readout and resulting low bandwidth in the phase direction.

The critical technical considerations of DWI for body applications are:
  • Motion compensation
  • Fat suppression
  • Parameter optimization
  • Adjustment of b-value

1. Motion compensation

A variety of motion correction techniques can be used in body DWI to minimize the effects of breathing. Each method will have associated advantages and limitations that make each more or less attractive based on the clinical requirements.

Breath hold
This option uses a simple breath hold to control artifacts from breathing. The DWI_BB_BH sequence in the Philips liver “Specials” folder is optimized for a single breath hold, single b-value (20) scan in approximately 20 seconds. Given that SNR decreases with higher b-values, breath hold scans will in practice be limited to relatively low b-values.

Navigator based triggering
This option uses a 2D navigator for breathing artifact reduction. Navigator uses the right hemi- diaphragm as the physiologic trigger and does not require the respiratory gating sensor.
Position the navigator on the dome of the liver using a free-breathing sagittal and coronal scout to include 1/3 lung and 2/3 liver on an expiration breath hold survey:
 Optimal navigator positioning for liver DWI <br>
Optimal navigator positioning for liver DWI
Respiratory triggering using sensor
This option uses respiratory triggering and requires use of the respiratory sensor. The motion of the abdominal wall is used as the physiologic trigger. The trigger delay should be optimized based on the respiratory rate to assure data collection occurs in the expiration phase:

Free breathing without gating
A Motion tab NSA value > 4 can be used to reduce motion artifacts from breathing as an alternative to navigator or respiratory triggering. However, some unsharpness will still be seen due to through-plane motion of the abdominal organs.

2. Fat suppression

Complete fat suppression is necessary to eliminate artifacts from the increased phase shifts between fat and water. A frequency selective spectral fat suppression pulse (SPIR/SPAIR) with volume shimming is used to eliminate fat signal. SPIR is typically more time efficient and lower SAR relative to SPAIR. However, SPAIR can be a good alternative for larger FOVs to yield more uniform fat suppression with a modest increase in scan time.

As with any spectral suppression technique, fat suppression non-uniformity is common in areas with magnetic susceptibility issues. Air/soft tissue interfaces can be especially problematic in body imaging; consequently, proper placement of a volume shim will be critical to spectral fat suppression. For best results, minimize air from the lung bases within the shim volume as shown:

DWIBS (Diffusion Weighted Whole Body Imaging with Background Body Signal Suppression) uses inversion recovery (IR) based fat suppression much like conventional IR sequences. The level of fat suppression is controlled by manipulation of the parameter TI based on the scanner field strength.

Inversion recovery based suppression typically results in longer scan times due to the nature of IR-based pulse sequences. However, IR-based fat suppression is less sensitive to B0 inhomogeneity and can be a useful alternative for improving fat suppression if SPIR or SPAIR yields unsatisfactory results.

For 3T systems, the Geometry tab parameter Gradient Reversal Offset Suppression (LIPO) is used to improve fat suppression for DWI. LIPO reverses the polarity of the selection gradient during the 90° excitation pulse compared to the refocusing pulse so only the water signals in the slice are excited by both RF pulses. LIPO is compatible with DWI using SPIR, SPAIR, and IR based fat suppression (DWIBS), but cannot be used with DWI ZOOM.

 3T DWI without (left) and with LIPO (right)<br><br><br>
3T DWI without (left) and with LIPO (right)

3. Parameter optimization

Proper optimization of body EPI sequences requires achieving the lowest possible water/fat shift (WFS) and EPI factor. Both WFS and EPI factor are shown on the Info page. Parameter adjustments cause the EPI factor and WFS to increase or decrease. In general, best image quality is achieved with the lowest practical WFS and EPI factor.

The parameters that contribute to reducing WFS and/or EPI factor are as follows:

Sensitivity Encoding (SENSE) is a parallel imaging technique used to control susceptibility artifacts common to EPI sequences. Phased array coils must be used and a reference scan performed whenever SENSE is enabled. As SENSE P reduction increases, EPI factor and WFS decreases to decrease susceptibility artifacts. Lower SNR and potential for SENSE artifacts are potential consequences of excessive P reduction.

Use a sufficiently large FOV to cover the anatomy without aliasing to minimize SENSE artifacts, or use fold-over suppression when a smaller FOV is desired. P reduction factor = 2 is a good starting point for balancing artifact reduction with SNR loss.

With the Ingenia system, higher P reduction factors (> 3) are possible in body DWI for further WFS/EPI factor reduction without artifact when:
  • Anterior/posterior coil combination used
  • Foldover direction = RL
  • Patient in arms up position


Comparison of arms down vs arms up with high SENSE P reduction. Note the difference in distortion of left lobe of liver

Field of View (FOV)
EPI factor and WFS increases as FOV increases. Note that flexible oversampling changes the FOV in the phase direction; consequently, EPI factor and WFS increases as flexible oversampling increases. Note that use of flexible oversampling effectively increases FOV in the phase direction and consequently affects WFS and EPI factor.

Voxel size
As in-plane voxel size increases, EPI factor and WFS decreases. Given the impact on in-plane spatial resolution, voxel size adjustment must balance artifact reduction with image sharpness.
Water fat shift (WFS)
WFS (in pixels) should always be set to Minimum to minimize artifacts from residual fat appearing in the area of interest. Note that many parameter changes will have a significant impact on the actual WFS value seen on the info page. The info page should always be checked after a parameter change to assure WFS did not increase significantly.
PNS mode / Gradient mode
Adjustment of PNS mode and/or Gradient mode affect gradient performance and can have an impact on WFS. Adjustment of PNS mode must balance WFS reduction with the potential for complaints of peripheral nerve stimulation.

Average high b
Signal to noise (SNR) and scan time is traditionally managed by the Motion tab parameter Number of signal averages (NSA). For DWI, a more time-efficient way to manage scan time and SNR by b-value is available with the Contrast tab parameter Average high b. With Average high b, a specific NSA value can be assigned to a specific b-value, whereas the Motion tab parameter NSA increases NSA to all b-values in the scan. This can be very useful in body applications like prostate DWI where three or more b-values are acquired.

If Average high b = yes, then each b-value is scanned with additional NSA according to the b-value:
For example, a DWI scan with b-values 0, 500, 1000 and NSA=2 with Average high b = yes increases NSA each b-value as follows:
  • b0: NSA = 2 (Motion tab) x1 (Average high b) = total NSA 2
  • b500: NSA = 2 (Motion tab) x2 (Average high b) = total NSA 4
  • b1000: NSA = 2 (Motion tab) x3 (Average high b) = total NSA 6

If Average high b = user defined, additional NSA can be applied to specific b-values without affecting other b-values, thus decreasing scan time. Since SNR decreases as b-value increases, higher NSA values can be targeted specifically to the higher b-values while lower NSA values are sufficient for lower b-values. Since NSA can be managed by b-value with user defined Average high b, the Motion tab parameter NSA can be set to 1.
To change NSA for specific b-values with the user defined average high b option, on the contrast tab click the drop down triangle symbol next to b-factor averages and enter the desired NSA for each b-value. In general, higher b-values need more NSA to preserve SNR, whereas lower b-values can have a lower NSA.

In summary, best image quality with body DWI are seen with the following parameter settings: 


  • WFS = minimum
  • Gradient mode = maximum
  • PNS mode = high
  • Highest practical SENSE P reduction
  • Smallest possible FOV
  • Balanced in-plane spatial resolution requirements vs. EPI distortions (EPI factor)
  • Average high b = user defined to efficiently manage NSA for each b-value

4. Adjustment of b-value

The b-value parameter adjusts the strength, duration and time of the diffusion gradients. Unlike brain imaging where b-value is well established (b=1000 s/mm2), in body imaging b-value requires optimization based on body part, pathology, and radiologist preference. For example, a b-value of 600 is a good starting point in liver imaging to balance lesion characterization and SNR. In other body applications, a b-value other than 600 may be desired based on the pathology. As always, consult the radiologist for the specific b-value(s) to be scanned.

A minimum of two b-values must be acquired to post process ADC maps in the Diffusion package. Typically, the first b-value is very low (0-100) and the second (and subsequent) is tailored to the pathology. ADC maps are generated on the scanner using the Diffusion software, which can be saved in the ExamCard as an automated in-line post-processing step.

Multiple b-values can be acquired in a single pulse sequence; however, scan time increases as number of b-values increases. SNR decreases as b-value increases and commonly requires a higher NSA value to regain signal. In this case, motion compensation techniques can be used to eliminate breathing motion with longer scan times. 
Prostate DWI showing effect of an increase in b-value on SNR

Examples of b-value optimization

Example 1: DWI_3b_RT (Ingenia 1.5T, Philips/Abdomen/Liver/Diffusion folder, Release 5.4 SW)

In this example, the DWI_3b_RT is optimized for b values 0, 100, and 650. On the Motion tab, change NSA = 1 and use the parameter b-factor averages to assign NSA to each b-value.

  1. On the Contrast tab, click the b-factors drop down triangle.
  2. For the third b-value, change the b-value from 800 to 650.
  3. Change parameter average high b from no to user defined.
  4. Click the b-factor averages drop down triangle.
  5. Change the number of averages as follows: b0 = 2, b100 = 2, b650 = 4

Note that four values are displayed even though three b-values are to be scanned. The default value for the fourth will be zero (0) in this scan.

The default scan used NSA = 4 (Motion tab) and average high b = no, which means each b-value would receive NSA = 4. With user defined average high b, lower NSA values can be assigned to lower b-values to save time.
Example 2: DWI_5b (Ingenia 3T, Philips/Pelvis/Prostate folder, Release 5.4 SW)
In this example, the DWI_5b is optimized for b values 0, 100, 500, 1000, and 1500. On the Motion tab, change NSA = 1 and use the parameter b-factor averages to assign NSA to each b-value.

  1. On the Contrast tab, click the b-factors drop down triangle.
  2. Enter b-values (in order): 1= 0, 2 = 100, 3 = 500, 4 = 1000 and 5 = 1500 (nothing is entered for line 6 since only 5 b-values will be scanned).
  3. Change parameter average high b from no to user defined.
  4. Change the number of averages as follows: b0 = 2, b100 = 2, b500 = 4, b1000 = 12, b1500 = 16

In the case of prostate DWI, b values > 1000 require high NSA values for adequate SNR. In this “all in one” example, the scan time will be over eight minutes.
Since Release 5.2 software, a b0 image is no longer required for the Diffusion package to accept diffusion data for ADC map calculation. In prostate DWI, the b0 scan is commonly not included in the ADC map calculation to minimize an artificial increase in ADC value called perfusion artifact. If the b0 is not used for the ADC map calculation, it can be removed from the scan and only b-values of 100, 500, 1000, and 1500 (DWI_4b) acquired: 

  1. On the Contrast tab, change nr of b-factors to 4.
  2. Click the b-factors drop down triangle.
  3. Enter b-values (in order): 1 = 100, 2 = 500, 3 = 1000, 4 = 1500.
  4. Change parameter average high b from no to user defined.
  5. Change the number of averages as follows: b100 = 2, b500 = 4, b1000 = 12, b1500 = 16.
In prostate imaging, the b1500 is used for lesion detection and may not be desirable to include it in the ADC calculation. If the b1500 image is not required for the ADC map calculation, the b1500 scan can be performed as a separate scan. In this case, a split b-value method is used for two scans (DWI_3b and DWI_b1500) where the b-values used for the ADC calculation (100, 500, 1000) are in one scan and the b value for lesion detection (1500) in the second scan.

For DWI_3b, change NSA = 1 on the Motion tab and use the parameter b-factor averages to assign NSA to each b-value.

  1. On the Contrast tab, change nr of b-factors to 3.
  2. Click the b-factors drop down triangle.
  3. Enter b-values (in order): 1 = 100, 2 = 500, 3 = 1000 (nothing is entered for line 4 since only 3 b-values will be scanned).
  4. Change parameter average high b from no to user defined.
  5. Change the number of averages as follows: b100 = 2, b500 = 4, b1000 = 12.

For DWI_b1500, change Motion tab parameter NSA = 16 and Contrast tab parameter average high b to no.

Compared to the all-in-one DWI method, the split b-value method improves patient compliance as the patient can much better tolerate two separate scans of four minutes or less with a pause between scans rather than a single scan of over eight minutes. Repeating a single scan of four minutes in case of patient motion is more tolerable than repeating a single eight minute scan as with the all-in-one method.

5. ADC map calculations in body DWI

A minimum of two b-values in a DWI scan is required for the Diffusion package and generating ADC maps. The Diffusion Registration feature is intended for brain DWI and not recommended for use in body DWI applications.

Prior to Release 5.2 SW, a b0 scan is required for the Diffusion package to accept the data. In body DWI, the b0 image may create an undesirable increase in ADC value due to capillary perfusion of water molecules. While the b0 image could easily excluded from the ADC calculation, the scan time to acquire the b0 images was essentially wasted. In Release 5.2 SW and later, the b0 scan is no longer required for the Diffusion package to accept the data. This can be useful in body DWI applications like prostate where the b0 information may not be desirable for inclusion in the ADC map.

In this example, two ADC maps are created in the Diffusion package from the same DWI scan with acquired b-values 0, 100, and 1000. The first ADC map (left) includes all three b-values, whereas the second ADC map (right) includes only b-values 100 and 1000 and excludes b0. An ADC measurement by ROI shows an increase in the reported ADC value with all three b-values compared to the ADC map where b0 was excluded. 
To exclude b-values from an ADC calculation: 

  1. In the Diffusion package, click the Select b-values button.
  2. Uncheck the b-values to be excluded from the ADC calculation.
  3. Click OK and generate the ADC map. 




The need for different (and multiple) b-values for various body applications makes DWI optimization more challenging compared to DWI in the brain. As discussed, a variety of tools are available to optimize SNR, EPI distortion, WFS, and motion in body DWI applications. While some options like b-factor averages are specific to Release 5 software, the suggestions in this article can be generally applied across different software levels.

This content has been made possible by NetForum Community.
Share this on: Share your link in twitter Share your link in facebook Share your link on LinkedIn Print Rate this article: Log in to vote

Jan 29, 2019

Rate this:
Log in to vote

Application Tip
Achieva 1.5T, Achieva 3.0T, Ambition, Elition, Ingenia 1.5T, Ingenia 3.0T, Intera 1.5T, Intera 3.0T, Multiva 1.5T, Prodiva 1.5T
Release 5
Body, b-values, Diffusion, DWI, DWIBS, LIPO

Clinical News
Best Practices
Case Studies
Publications and Abstracts
White Papers
Web seminars and Presentations
Application Tips and FAQ
Try an Application
Business News
Case Studies
White Papers
Web Seminars and Presentations
Utilization Services
Contributing Professionals
Contributing Institutions
Become a Contributor