In order to bring you the best possible user experience, this site uses Javascript. If you are seeing this message, it is likely that the Javascript option in your browser is disabled. For optimal viewing of this site, please ensure that Javascript is enabled for your browser.
Did you know that your browser is out of date? To get the best experience using our website we recommend that you upgrade to a newer version. Learn more.

We use cookies to optimise the design of this website and make continuous improvement. By continuing your visit, you consent to the use of cookies. Learn more

MRI evaluation of aortic stenosis: flow evaluation

Author: Cemil Izgi


Cardiovascular magnetic resonance (CMR) can provide reliable information on flow and velocity data through any stenotic or regurgitant valve, very much like the Doppler study of echocardiography. The imaging technique used is phase contrast (PC) velocity mapping and relies on differential build-up of phase in flowing protons in comparison to protons in the static tissues. The technique is widely used and validated in the assessment of valvular heart diseases. For the assessment of aortic stenosis main use of PC imaging is to derive maximum velocity (Vmax) of the aortic stenosis jet.

Phase Contrast CMR Imaging (also known as flow mapping or velocity mapping)(Figure-1, A-B)

Phase refers to angular position of a proton`s spinning vector. For PC imaging not the magnitude of the protons` spinning vector but their phase (angle) data are used. A bipolar gradient pulse pair (a pair of gradients with same field strength but opposite directions) is applied in the direction of flow. The phase difference induced by the first bipolar gradient in the static tissue is completely reversed by the second gradient (which is of same strength but opposite direction). Therefore protons in the static tissue have a net phase of zero after the bipolar gradient is applied. However since protons of flowing blood are moving along the gradient they are exposed to different gradient strengths during acquisition and their net phase is not zero. The net phase acquired by the flowing protons is proportional to their velocity and PC CMR imaging uses this principle to derive velocity of flow.

When no magnetic gradient is applied the magnetic field is homogenous and therefore all the protons of the flowing blood and the static tissue are spinning in-phase (no angular difference between them). Immediately after the fist gradient, a second gradient which is of the same magnitude as the first one but in opposite direction is applied (second half of the bipolar gradient pulse pair).
Figure-1, A - When no magnetic gradient is applied the magnetic field is homogenous and therefore all the protons of the flowing blood and the static tissue are spinning in-phase (no angular difference between them). Assume that we want to measure the velocity of blood flow while it moves between the two pink lines. Figure-1, B - Immediately after the fist gradient, a second gradient which is of the same magnitude as the first one but in opposite direction is applied (second half of the bipolar gradient pulse pair). The protons in points b and c are exposed to same gradient strength as in the first half, but in opposite direction. This gradient returns all the protons in the static tissue back to their phase at the start, they are now in-phase again. However the proton in the flowing blood has now moved from a low magnetic gradient strength location to a higher strength location and it has a net phase difference (Ø) compared to the protons of the static tissue. The amount of this phase difference is proportional to how fast the protons of flowing blood move in the magnetic field gradient and is used to calculate the velocity of flow.


The velocity is displayed on the gray scale for each of the pixels in PC image with highest intensities on the two ends of the scale reflecting the encoded maximum velocities in opposite directions (Figure-2). The PC image is generally paired with the corresponding magnitude cine images.

The net phase difference of the flowing protons is proportional to the velocity of the flow as the faster moving protons will gain more phase difference. Velocity can be encoded in three directions (x,y,z coordinates) of any imaging plane.
Figure 2 - The net phase difference of the flowing protons is proportional to the velocity of the flow as the faster moving protons will gain more phase difference. Protons moving in the opposite of the encoding direction will gain negative phase difference, the degree of which is again proportional to their velocity. Sensitivity of the velocity encoding (velocity corresponding to maximum phase difference; +/- 180°) is user selected. Velocities above the encoding velocity sensitivity are aliased (e.g., a phase difference of +270° will be read as -90°). Figure 3 - Velocity can be encoded in three directions (x,y,z coordinates) of any imaging plane. Velocity is encoded in either of the two dimensions of the in- plane PC images. Through plane images are obtained by obtaining an image slice perpendicular to flow. Through plane images show a flow throughout its borders and can be used to derive flow through a vessel. A region of interest (e.g., borders of a vessel) is drawn by the user in every frame of the through plane cine images in a cardiac cycle. Within this region of interest velocity and area of every pixel is known and all flow related data (flow and maximum/mean velocity curves) can be derived by use of a software.

PC velocity mapping can be applied for within the plane (in-plane) or through the plane of imaging (through plane) (Figure-3). When through plane phase mapping is used velocities of all the pixels in a region of interest can be derived and flow through the region of interest can be quantified. For AS, however, the main index of severity to be derived from phase mapping is Vmax of flow through the stenotic valve.

Acquisition of images: planning

First, the stenotic aortic valve is visualised in the left ventricluar outflow tract (LVOT) long axis and coronal cine images. The jet of the stenotic valve can be seen in early systole. In- plane phase contrast images are easily planned from these images. Simply, the systolic frame which shows the aortic stenosis jet best is selected (Figure-4, A). All needed is to rotate the field of view to ensure that the aortic stenosis jet direction is parallel to the velocity encoding direction and then PC sequence is run (Figure-4, B). An appropriate Venc should be selected (please see tips and tricks below).
Figure 4,A. Planning of PC images Figure 4A. Planning of PC images
Systolic Frame.
Figure-4,A. - Planning of PC images.From the LVOT long axis and coronal cine images systolic frames which best show the aortic stenosis jet are selected.
Figure 4B. In plane PC image
Figure-4, B
It is important to make sure that the flow direction is as close to velocity encoding direction as possible. To ensure this the imaging plane is rotated so that the direction of aortic stenosis jet is parallel to the side of the imaging frame in the direction of velocity encoding. Care should be taken to adjust image size to avoid wrap.

For through plane phase contrast images a slice is selected perpendicular to the jet. Either the cine images or the in-plane phase contrast images can be used to plan the location of the through plane image The maximum velocity of the aortic stenosis jet is above the level of the valve level so the slice should be arranged accordingly (Figure-4, C). To make sure that the peak velocity along the jet is acquired it may be better to acquire 2 contiguous and overlapping through plane slices above the level of valve (slice thickness 8mm and interval 5mm). As with the in-plane imaging an appropriate Venc selection is essential.

Most of studies and centers use 2D TEE measurement of aortic valve annulus diameter for TAVI sizing
  Venc 400cm/s.
Venc 450cm/s
Figure-4, C - A through plane image is acquired perpendicular to the aortic flow jet. The location of the image slice is adjusted in the LVOT cine views with the aim of acquiring an image through which the jet velocity is maximum (upper row). The LVOT in plane PC images are also of help when planning. After planning the most appropriate image position, a Venc value is set depending on the expected velocity of the jet. If aliasing occurs (as in this case), then the sequence is run once again by increasing the Venc. To make sure that peak velocity along the jet is imaged, it may be better to acquire contigous and overlapping slices above the level of valve and each of these images are then checked for the maximum velocity.

Analysis of phase contrast flow images

Through plane phase contrast images should be preferred to in-plane images for estimation of peak velocity (see tips and tricks section).

Conventionally, contours of the aorta are traced in all phases of the cardiac cycle in through plane images to derive a velocity time curve for the aortic flow. From this curve maximum velocity and all the other parameters of flow are derived (Figure-5, A). However for the case of aortic stenosis a practical approach is to identify the brightest (or darkest) pixel in the PC flow images as this pixel will be encoding the highest velocity (Vmax) of aortic stenosis (Figure-5, B).

Figure-5, A
Figure-5, A.


Figure-5, B.
Figure-5, B. - An easier way to derive Vmax from PC images is by finding the highest velocity pixel of aortic flow. This can be performed by adjusting the contrast settings to highlight the brightest (or darkest) pixel of the aortic flow in cine PC images. The contrast window is narrowed to the level that the image pixels are either white or black depending on the threshold level set. This helps to point out the pixel with highest velocity and intensity of this pixel is Vmax of aortic stenosis.

Tips and tricks

PC imaging technique utilizes a gradient echo sequence. The images can be acquired during breath-holding or free breathing and either by retrospective or prospective gating. Generally breatholding acquisitions are preferred as these are shorter. In prospective gating the last part of diastole is not acquired and is missing from the images of a cardiac cycle. This part of the cardiac cycle is important for lesion such as aortic regurgitation since aortic regurgitation occurs in diastole. However, in aortic stenosis main data needed from flow imaging is the maximum flow velocity (Vmax) which occurs in systole. Therefore, both retrospective and prospective gating are reliable for estimating Vmax in aortic stenosis.

Venc setting:

Venc is user selected and should be slightly higher than the actual velocity. If it is lower than the actual velocity aliasing occurs and if it is largely higher than the actual velocity the sensitivity to adequately detect maximum velocity decreases. Ideally it should not be 20% higher than the actual velocity. Imaging should start with a Venc according to the estimated maximum velocity and then adjusted accordingly.

Imaging plane alignment:

The direction of the flow should ideally be parallel to velocity encoding direction or at most be within an angle of 10-15°.

Partial volume averaging:

The velocity of a pixel in the image is derived from mean signal received from its respective voxel. For PC imaging if a voxel includes areas of high and low velocity flow, the mean velocity of the voxel will be lower than the highest velocity. The larger dimension of the voxel is the slice thickness (typically 8mm). For the in-plane image slice thickness will extend through both the high velocity core of the AS jet and also the lower velocity peripheral regions. However for the through plane image the slice thickness will most probably extend through the high velocity central core of the jet. Therefore through plane PC images are more reliable in estimating Vmax in aortic stenosis.

Temporal resolution:

With the current scanners, for the PC images a frame rate of 25-30 phases/cycle can be achieved which is adequate for estimating Vmax. If the temporal resolution is less then it is possible to miss the highest velocity of aortic stenosis jet during cardiac cycle and therefore underestimate Vmax.

Phase offset errors:

Due to magnetic field inhomogeneities small phase errors may occur which leads to errors in estimated flow velocities. These are more of a problem for flow quatification since even small errors may translate into large differences in flow when the erroneous velocities are multiplied by the area. However, for determination of Vmax small phase offset errors are acceptable.

Using lowest possible echo time:

Spins of flowing blood dephase quickly especially in the case of turbulance and this causes loss of signal. Using the shortest echo time possible for PC imaging is therefore recommended.

Unresolved issues

Some unresolved or controversial issues may be worth mentioning:

Measuring area of aortic jet to estimate aortic valve area:

One may suggest measuring the area of aortic flow jet in through plane PC images to estimate the aortic valve area. This may look similar to measuring jet thickness or vena contracta in a valvular lesion with colour Doppler echocardiography. Although they may look similar, the origin of flow signal in PC imaging is different than that of colour Doppler. And among many other factors, turbulence related dephasing affects the area of aortic stenosis jet and this method is not recommended for estimating aortic valve area.

Aortic valve area by continuity equation based on CMR phase contrast imaging:

CMR has the advantage of estimating flow in any plane and through any cardiac chamber. It is possible to calculate aortic valve area by acquiring two flow images one through the LVOT and other above the aortic valve level. All the information needed to calculate aortic valve area by continuity equation can be derived from these images (velocity time integral of aortic and LVOT flow and LVOT cross sectional area). A study has shown validity of this approach to calculate the aortic valve area. However, measurement of aortic valve area directly by planimetry is a much simpler method which have been widely used and validated.

Assessment of low flow, low gradient severe aortic stenosis:

The recent concept of parodoxical low flow, low gradient severe aortic stenosis suggests that substantial proportion of patients with severe aortic stenosis (based on valve area criteria) may have lower gradients and peak velocities due to limited aortic flow. These patients in fact may be in a more advanced phase of severe aortic stenosis. The major criteria for identification of these patients is first to estimate aortic flow rate, as a low-flow state is defined as stroke volume index of <35ml/m2. With CMR imaging, left ventricular stroke volume can easily be derived from the aortic flow data as well as from the volumetric analysis of the left ventricle. Usefulness of CMR derived stroke volume index in defining low flow, low gradient severe aortic stenosi has not been demonstrated yet but may prove valid in future studies.