Phase-Contrast MRI — Velocity Encoding in Practice, VENC Selection, Morphological Output (PC-MRA / MRV), Quantitative Output (2D Cine Flow, 4D Flow), Phase Errors and Evidence
Linked parent page: Flow, Motion, and Diffusion Physics
Phase-Contrast MRI — Velocity Encoding in Practice, VENC Selection, Morphological Output (PC-MRA / MRV), Quantitative Output (2D Cine Flow, 4D Flow), Phase Errors and Evidence
MRIninja Knowledge Base | Sequence/Technique Child Page Builds on: Flow, Motion and Diffusion Physics (Section 3.2) · MRI Sequences Overview (Section 8.7, VENC) Version 1.0 — October 2026
1. Introduction and Scope
The physical principle of phase-contrast (PC) imaging — a bipolar gradient that leaves stationary spins unchanged and gives moving spins a phase proportional to velocity — is established on the Flow, Motion and Diffusion Physics page, and the basic definition of VENC on the MRI Sequences Overview. This page assumes both and does not re-derive them. It addresses what they do not: how a PC acquisition is built and used, and how its two distinct outputs differ.
- A morphological output (phase-contrast angiography and venography, PC-MRA / MRV): the vessel is displayed, with flow direction, and the quality criterion is vessel conspicuity.
- A quantitative output (2D cine flow measurement and 4D flow): the velocity itself is the result, and the quality criterion is accuracy, with a defined error budget.
The same acquisition parameters have opposite consequences in the two uses (Sections 3, 4 and 5), which is the reason for treating them together on one page. For the inflow-based alternatives, see the triggered non-contrast MRA and inflow-IR bSSFP pages.
Information status. The quantitative branch rests on society consensus documents. The morphological branch rests on educational and review sources, without an angiography-specific guideline (Section 10). Several primary sources were available only as abstracts or figure legends, which is stated in the bibliography. Numerical values are quoted with their source and are not recommended settings.
2. How the Data Are Formed in Practice
Two acquisitions are made that differ only in their flow-sensitising bipolar gradients; subtraction of the two removes phase shifts caused by the imaging gradients themselves [5]. The review of Wymer and colleagues adds that stationary-tissue signal cancels in the subtraction and that "the contribution of susceptibility to the phase shift also will be removed" [4]. Two image types result: the magnitude image, used for anatomical orientation, and the velocity (phase-difference) image, which carries the flow information [5].
| Mode | Encoding | Typical product | Source |
|---|---|---|---|
| 2D cine PC | One direction ("through-plane"), motion-encoded cine gradient-echo, gated | Flow curve through a single plane | [2] |
| 3D PC (angiographic) | Velocity encoding in a volume | PC-MRA / MRV; the mrimaster resource states that PC can be used for 2D and 3D imaging with brain MRV as a 3D example | [8] |
| 4D flow | Three-directional velocity encoding in a volume, resolved over the cardiac cycle | Volume allowing retrospective flow calculation through any plane | [1,3] |
A 4D flow acquisition is described by the 2015 consensus as one volume that is "easy to place" and that allows flow through any plane within it to be calculated retrospectively, with typical spatial resolution of 1.5–3 mm, temporal resolution of 30–40 ms and scan times of 5 to 25 minutes [1].
3. VENC — One Parameter, Two Opposite Optimisation Rules
3.1 What VENC Controls
VENC is "inversely related to the first moment of the bipolar gradient waveform": stronger or longer flow-encoding gradients produce more phase shift and therefore a lower VENC [9]. The maximum correctly measured velocity is inversely proportional to the change in magnetic moment [4]. Phase angles beyond ±180° wrap around and are misread (aliasing) [4]; for example, +75 cm/s with a VENC of 50 cm/s is displayed as −25 cm/s [9].
3.2 Selection Rules in the Sources
| Use | Rule stated | Source |
|---|---|---|
| General statement | Choose about 20–25% above the maximum expected velocity; if the optimum is unknown, repeat with two or three settings | [9] |
| 2D flow quantification (SCMR 2020) | "The lowest available velocity without aliasing should be used"; check phase-difference images after each scan and increase VENC if aliasing appears; a velocity scout can help | [2] |
| 4D flow (2023 update) | VENC "close to the maximum velocity (< 25% above)", "as low as possible to keep adequate VNR and improve accuracy while avoiding aliasing"; if stenosis is suspected and no prior study exists, an initial VENC of 250 cm/s may be considered; regions with incorrigible aliasing should not be analysed | [3] |
| Error behaviour (Lotz 2002) | Start with a high VENC and repeat with an adapted value if peak velocity is needed; a VENC more than three times the in-vessel velocity shifts peak-velocity estimates by more than 10% while flow estimates are largely preserved, and noise increases; aliasing below about one-third of Vmax cannot be corrected manually or by software | [5] |
| PC-MRA (educational resources) | Higher VENC for arteries (40–70 cm/s) and lower for veins (10–20 cm/s); too high a VENC leaves slowly flowing vessels poorly depicted | [8,9] |
Two points follow. First, for quantification the aim is the lowest VENC without aliasing, because a higher VENC lowers the velocity-to-noise ratio [2,3,9]. Second, for morphology the VENC is a contrast control: it selects which vessels are displayed, and a value suited to arteries will under-represent veins [8,9]. Wymer and colleagues likewise state that changing the VENC facilitates imaging of the arterial or the venous system [4]. The numerical ranges differ between sources and depend on the application: the VENC values quoted for PC-MRA in the educational resource [8] are lower than the expected-velocity ranges (arteries about 100–200 cm/s, veins about 20–40 cm/s) given on the Sequences Overview page; the two kinds of value are not the same quantity, and the table above shows how the rule applies in each use. No retrieved source defines a single standard.
4. Morphological Output — PC-MRA and MRV
4.1 What the Image Shows
A phase-difference image has signal "linearly proportional to the velocity of the spins", with the sign carrying direction: spins moving in one direction appear bright and those moving in the opposite direction dark [8]. Background suppression arises from the bipolar gradient diminishing stationary-tissue signal [8]. This makes PC-MRA a flow-direction-aware, velocity-weighted display, unlike the saturation- or inversion-based inflow contrast of the techniques on the triggered MRA pages. The Sequences Overview lists venous sinus assessment and idiopathic intracranial hypertension among its indications for PC-MRA.
4.2 Practical Points
- VENC tuned to the vessel class (Section 3.2).
- Several VENC values can be used in different scans to emphasise different vessels [8].
- Complex or turbulent flow. Reasoning, not a quoted statement: the intravoxel dephasing described on the flow physics page (Section 4.2) reduces signal behind stenoses and in aneurysms, so signal loss on PC-MRA in such places is not specific for occlusion.
No retrieved source gives acquisition times or diagnostic accuracy for PC-MRA relative to TOF-MRA or contrast-enhanced MRA, and none is cited here for those comparisons (Section 10).
5. Quantitative Output — 2D Cine Flow Measurement
5.1 Technique as Recommended in the SCMR Protocols
| Item | Recommendation |
|---|---|
| Sequence | One-direction ("through-plane") motion-encoded cine gradient-echo |
| Plane | Centred on the vessel and at scanner isocentre; "aligned orthogonally to the expected main blood flow direction in two spatial directions" |
| Slice and in-plane resolution | 5–8 mm slices; in-plane resolution at least 1/10 of the vessel diameter |
| Temporal resolution | ≤ 50 ms between phases, measured from the frames actually acquired (interpolated frames only make resolution appear better) |
| Gating | Retrospective gating covers the whole cycle but "may obscure inaccuracies related to arrhythmia"; segmented acquisition over several heartbeats fits a breath-hold; navigator-based or free-breathing multi-average approaches are alternatives |
| Echo time | As short as possible, particularly at stenoses |
| Check | Review phase-difference images for aliasing after each acquisition |
The SCMR document also states that, if available, a 4D flow acquisition should be considered [2]. All items in this table are from the SCMR protocols [2]; the document is cardiovascular, and its applicability to other territories (CSF, cerebral vessels) is not established by it.
5.2 Error Budget from the Lotz Education Exhibit
| Error source | Finding reported |
|---|---|
| Plane deviation | More than 15° from perpendicular causes more than 10% error in both peak velocity and flow; "Flow measurements are most precise if the imaging plane is perpendicular to the vessel of interest" |
| Temporal resolution | Reconstructing more than 16 frames per cardiac cycle did not improve accuracy in that study; the authors chose 30 as their standard |
| Spatial resolution | Flow is more sensitive to poor spatial resolution than peak velocity is |
| VENC | See Section 3.2 |
| Background (phase offset) subtraction | Small changes in the correction can cause large flow differences, so it should be used sparingly |
The authors state that "with phase-contrast imaging, the overall error in flow measurement can be reduced to less than 10%" [5]. These data are from an educational exhibit read through its abstract and figure legends and do not replace a validation study; the SCMR value of ≤ 50 ms and the 30 frames of Lotz refer to different parameters and are not directly comparable.
5.3 Derived Quantities Reported
The Wymer review lists cardiac valve flow, with use to "assess the severity of valvular disease and quantify the shunt fraction", CSF flow, vascular imaging, 4D flow and MR elastography as applications [4]. For CSF, the review's figure legend gives an aqueductal CSF stroke volume "greater than 50 µL" as supporting a diagnosis of normal-pressure hydrocephalus [4]; this is stated in a single figure legend and is not a validated threshold on this page. The Leiden educational syllabus lists transstenotic pressure drop by the Bernoulli equation, kinetic energy and wall shear stress as derived measures, with misalignment, phase aliasing and phase offset errors as error sources [6].
6. Phase Offset Errors — The Principal Threat to Quantification
Stationary tissue should show zero velocity. In practice the phase image contains a spatially varying offset related to eddy currents and concomitant gradient fields, which must be corrected using static tissue or a static phantom [3,7]. Gatehouse and colleagues measured the uncorrected offsets on twelve 1.5 T systems from three vendors (four per vendor) using static gelatin phantoms, breath-hold retrospectively gated PC acquisitions and a velocity encoding range of 150 cm/s: the largest uncorrected offset in a 30 mm "great vessel" region ranged from 0.4 to 4.9 cm/s (mean 2.7 cm/s), and an offset of 0.6 cm/s was calculated to be barely acceptable, causing roughly 5% error in cardiac output and up to 10% in shunt measurement [7]. The authors concluded that all tested systems appeared to need post-acquisition correction and that the effectiveness of correction software remained to be tested on clinical flow acquisitions [7].
The 2023 consensus states that offset errors increase with distance from the magnet isocentre, that static-tissue interpolation can be applied during post-processing "with equivalent performance" to other correction, and that publications should describe the correction methods for eddy currents and for distortions from gradient field non-uniformity [3]. Gradient non-linearity and concomitant-field (Maxwell) terms are not discussed in detail in the retrieved passages and are not developed here. The phantom study was performed on 1.5 T systems of the 2000s and its figures should not be assumed to apply to current hardware.
7. 4D Flow
The 4D flow consensus documents (2015 and 2023 update) give the following recommendations and statements [1,3]:
| Topic | Statement |
|---|---|
| Spatial resolution | Isotropic voxels with at least 6 voxels across the vessel diameter; acquired and reconstructed resolution should both be stated [3] |
| Temporal resolution | Upper limit 50 ms (may be higher if the aim is visualisation only), aim 30 ms [3] |
| Scan time | Advisable to keep to 5–10 minutes; respiratory compensation, multi-VENC and a large field of view increase it [3] |
| Validation against 2D | Compare forward flow and peak velocity; ideally differences ≤ 5%; scan-rescan differences up to 10% acceptable [3] |
| Prognostic evidence | "All published prognostic values are based on 2D Flow CMR"; initially both 2D and 4D flow should be acquired and analysed in parallel [3] |
| Role | Flow quantification may take priority for shunt evaluation and valvular regurgitation; adjunctive role, for example in aortic aneurysm; qualitative analysis may help delineate stenosed vessels in congenital heart disease [3] |
| Parameters needing validation | Wall shear stress, pressure difference, turbulent kinetic energy and intracardiac flow components (2015) [1] |
| Limitations | Accuracy can be compromised in certain flow geometries, such as high-velocity jets and regions with substantial turbulent dephasing; irregular heartbeats are a challenge, and retrospective valve tracking in atrial fibrillation may underestimate flows [3] |
These documents concern cardiovascular MR. Their extension to intracranial vessels and CSF is not addressed in the retrieved passages.
8. Pitfalls and Failure Modes
| Problem | Mechanism or evidence | Response |
|---|---|---|
| Aliasing | VENC below the peak velocity; "pixelated focus of opposite signal intensity at the centre of the flow" at 200 cm/s that resolved at 250 cm/s in a clinical example [4] | Review the phase image; raise VENC; unrecoverable below about one-third of Vmax [5] |
| VENC too high | Loss of velocity-to-noise; slow vessels poorly depicted [9] | Lowest VENC without aliasing for quantification; matching VENC to vessel for morphology |
| Plane not perpendicular | More than 15° off gives more than 10% error [5] | Plan on orthogonal localisers in two directions [2] |
| Uncorrected phase offset | 0.4–4.9 cm/s on phantoms [7]; increases away from isocentre [3] | Correct; centre the vessel at isocentre [2,3] |
| Arrhythmia | Retrospective gating may hide inaccuracies [2]; irregular beats challenge 4D flow [3] | Check the gating signal; interpret with caution |
| Misleading temporal resolution | Interpolated frames [2] | State the acquired resolution |
| Signal loss in jets and turbulence | Intravoxel dephasing (Section 4.2 of the flow physics page); accuracy compromised [3] | Short TE [2]; do not analyse affected regions [3] |
9. What to State When Reporting
For the quantitative output, the 2023 consensus asks that methods describe the VENC strategy, acquired and reconstructed resolution, and the correction methods for eddy currents and gradient non-uniformity [3]. Reasoning, not a quoted statement: the same items — VENC, plane orientation relative to the vessel, gating mode and temporal resolution, and whether background correction was applied — are what a reader needs to judge a flow value in a clinical report; for the morphological output, the VENC used and the vessel class it targets are the corresponding items.
10. Evidence Gaps and Ongoing Debate
- No guideline specific to PC-MRA / MRV was found. The morphological section rests on educational resources [8,9] and a review [4]; acquisition time, accuracy and comparison with TOF or contrast-enhanced MRA are not supported by a retrieved source.
- The consensus documents are cardiovascular. Their extension to neuro-vascular and CSF applications is not established here.
- Sources read in part. The Wymer review, the Lotz exhibit and the 2015 4D flow consensus were read through abstracts and figure legends or page summaries, not in full text; the 2023 update and the SCMR 2020 protocols were read through page extracts.
- Phantom-based offset data. The Gatehouse figures come from static phantoms on 1.5 T systems and did not test correction software on clinical acquisitions.
- VENC ranges are inconsistent between sources and with the site's Sequences Overview page (Section 3.2); the overview page was not modified in this release.
- Multi-VENC, gradient non-linearity and concomitant-field correction are only touched on in the retrieved passages.
- The CSF stroke-volume threshold for normal-pressure hydrocephalus comes from a single figure legend and was not verified against primary studies.
11. Evidence-Based References
No official guideline specific to phase-contrast angiography was verified. References are listed by evidence tier and recency.
B. Consensus Statements and Society Protocols
C. Original Studies
D. Technical Reviews and Educational Sources
End of document — Phase-Contrast MRI — MRIninja v1.0 — October 2026
Related Protocols
Recent PubMed search for this protocol