Journal of Cardiovascular Magnetic
Comparison of strain-encoded cardiac MRI images with
k-space acquisition strategies
El-Sayed H Ibrahim*1, Wolfgang Rehwald2, Sven Zuehlsdorff2 and
Richard D White'
Address: 'University of Florida College of Medicine, Jacksonville, FL, USA and 2Siemens AG Healthcare Sector-Cardiovascular MR Research and
Development, Chicago, IL, USA
* Corresponding author
from 13th Annual SCMR Scientific Sessions
Phoenix, AZ, USA. 21-24 January 2010
Published: 21 January 2010
journal of Cardiovascular Magnetic Resonance 2010, I2(Suppl I):P93 doi:10.I 186/1532-429X- 12-SI-P93
This abstract is available from: http://jcmr-online.com/content/ 2/S I/P93
2010 Ibrahim et al; licensee BioMed Central Ltd.
Strain encoding(SENC) is a newly-developed MRI tech-
nique for measuring myocardial strain in the through-
plane direction . With conventional k-space acquisi-
tion, SENC requires two separate acquisitions to obtain
two sets of images (low-tune(LT) and high-tune(HT)),
which are combined to obtain the strain image. Non-Car-
tesian k-space trajectories are becoming more popular
because they allow for data undersampling with accepta-
ble image quality . In this work, SENC was combined
with undersampled radial k-space acquisition and inter-
leaved SENC tunings to reduce scan time to one breath-
hold. The results were compared to conventional SENC
Three volunteers were scanned on 3 T MRI system (Sie-
mens TIM-TRIO, Erlangen, Germany). Four SENC scans
were conducted in four separate breath-holds. Conven-
tional SENC pulse sequence was run in the first and sec-
ond scans: in the first scan, a set of cine LT images was
acquired with Cartesian k-space (no undersampling). The
second scan was the same as the first scan, but with HT
images. In the third scan, the modified SENC sequence
(Fig. 1) was used: interleaved LT and HT tunings with
radial undersampling (60%). The fourth scan was a repe-
tition of the third scan, but with Cartesian undersampling
instead of radial (same ratio), for comparison. In all scans,
ramped flip-angle was used to maintain constant signal
intensity. Pixel size ~1.5 mm2 and scan-time ~20 s.
The LT and HT images were combined in an interleaved
fashion (view-sharing) to result in strain images (about 26
phases). Signal-to-noise ratio(SNR) was measured in each
image by dividing mean myocardial signal intensity by
I .n I I I
l -n I t --------i1
Z-I. I Slc h "/ Ti
------u --- -M- -- -f -'
Modulation Demodlatlion ,
-- I I I I laq. I dlAcq. --
I I I
Modified SENC pulse sequence. The sequence consists
of two sections: modulation and imaging. Magnetization is
modulated using non-selective 900-900 RF pulses with z-gra-
dient in-between. During imaging, alternating low-tuning (LT)
and high-tuning (HT) demodulations are implemented. Radial
k-space reading is used with 60% undersampling. Imaging RF
pulses are ramped during the cardiac cycle to compensate
for modulation fading.
Page 1 of 2
(page number not for citation purposes)
Journal of Cardiovascular Magnetic Resonance 2010, 12(Suppl 1): P93
Short-axis SENC images showing longitudinal strain
at the same heart phase. Radial (left) and Cartesian
(right) acquisition. Arrows point to high-strain inferior and
low-strain anterior LV myocardium. Dotted arrows point to
RV. The radial image as acceptable image quality and acquired
in half the time as the Cartesian.
standard-deviation of background noise. Longitudinal
strain values were calculated at the same position and car-
diac-phase from corresponding Cartesian and radial
images. Bland-Altman analysis was conducted.
Fig. 2 shows the resulting strain images. The undersam-
pled radial images show similar image quality to conven-
tional Cartesian images. No streak artifacts were observed.
The undersampled Cartesian images(scan#4) resulted in
much deteriorated image quality. SNR was 20 and 16 for
Bland-Allman (Cartesian vs. Radial)
S4 --a------ ---- ---
2 -------*--------- ---------------- ------------------------
'S -2 ------ --b-i---- - ---- ---- ---------
4m .D"" ~. ~.~ --i-t-i----r--- T -''.----- .T--..
-4 -- ............... ......... i ...... -..---- .
12 14 16 18 20 22 24 26 28
Longitudinal strain average, -%
Bland-Altman plot for the correlation between myo-
cardial longitudinal strain from Cartesian and radial
SENC images. The plot shows no significant bias between
the two methods. All points lie within the 2SD margin.
Cartesian and radial images, respectively. Bland-Alt-
man(Fig. 3) shows no significant difference in strain
measurement between the two methods (differences lie
within + 2SD limits).
Discussion and conclusions
Radial k-space acquisition allows for SENC imaging in
one breath-hold with satisfactory image quality. Image
misregistration artifacts are minimized using the pro-
posed method. The high-field 3Tesla system compensates
for SNR loss associated with reduced data acquisition.
Employing the same undersampling ratio with Cartesian
acquisition results in deteriorated image quality and
introduces artifacts. Future studies will address imple-
menting 3D radial acquisition to improve SNR.
I. Osman : MRM 46:324-334.
2. Shankaranarayanan : Radiology 21:827-836.
3. Altman : Statistician 32:307-3 17.
Page 2 of 2
(page number not for citation purposes)
Publish with BioMed Central and every
scientist can read your work free of charge
"BioMed Central will be the most significant development for
disseminating the results of biomedical research in our lifetime."
Sir Paul Nurse, Cancer Research UK
Your research papers will be:
available free of charge to the entire biomedical community
peer reviewed and published immediately upon acceptance
cited in PubMed and archived on PubMed Central
yours you keep the copyright
Submit your manuscript here: BioMedcentral
hftp://jcmr-online. com/content/l 2/Sl/P93