Philips MSDE Black-Blood Preparation — MSDE Mode, Refocusing Pulses and Type, TEprep and Venc: Parameter-by-Parameter Use

Philips MSDE Black-Blood Preparation — MSDE Mode, Refocusing Pulses and Type, TEprep and Venc: Parameter-by-Parameter Use

MRIninja Knowledge Base | Technique Deep Dive (vendor parameter guide) Builds on: Vessel Wall Imaging — Intracranial Arteries (Section 10.1) · Vessel Wall Imaging — Universal Technique Master (Section 3.3) · Controlling Fluid Signal (Section 6) Version 1.0 — October 2026

1. Introduction and Scope

The need for a black-blood preparation and the three families of preparations (DIR, MSDE, DANTE) are on Vessel Wall Imaging — Universal Technique Master and Controlling Fluid Signal; the intracranial protocol that uses MSDE is on Vessel Wall Imaging — Intracranial Arteries. This page does not repeat them. It addresses the operator-level question: what each field in the MSDE block of the Philips black-blood (BB pulse) parameter list controls, what the literature says about its effect, and how the fields interact. The fields covered are those shown in the console list: BB pulse = MSDE; MSDE mode; refocusing pulses; refocusing type; TEprep (with its value in ms); Venc (RL, AP, FH). The MTC field listed below them is a separate parameter and is not treated here.

Information status — read this first. No Philips manual, application guide or publication that defines these console fields was found. What exists is (a) the original and improved MSDE literature, much of it from groups that used Philips scanners and in some cases Philips-affiliated authors, written in terms of the physics (number of 180° pulses, TEprep, first gradient moment m1), and (b) the console field names. The correspondence between the two is therefore an interpretation. Throughout, statements are tagged: [source] where a cited source says it; Reasoning where it is an inference or a calculation on this page. Section 9 lists what must be confirmed on the console.

2. What the Preparation Does and Where Each Field Acts

The preparation is a short module played before the TSE readout. A 90° pulse flips the magnetisation into the transverse plane; refocusing 180° pulses are interleaved with motion-sensitising gradients; a final 90° pulse returns the surviving component to the longitudinal axis, and residual transverse magnetisation is spoiled [4]. The gradients are balanced so that the total zeroth moment is zero, so stationary spins end in phase, whereas spins that move acquire a phase proportional to their velocity and the first moment m1 of the gradients [8, 10]. In the improved MSDE (iMSDE) the RF train is 90°–180°–180°–90° with the intervals TEprep/4, TEprep/2 and TEprep/4, TEprep being the time between the first and last 90° pulses, and with four motion-sensitising gradients of alternating polarity [8, 9].

The phase acquired by a spin moving at velocity v along the sensitised direction is φ = γ · m1 · v (γ = 2.675·10⁸ rad s⁻¹ T⁻¹). Reasoning: after the final 90° pulse a fraction cos φ of the magnetisation is returned to the longitudinal axis, so stationary tissue (φ = 0) is kept and spins with φ near π/2 are removed (Section 7). The same bipolar-gradient phase principle is the basis of phase-contrast imaging (Flow, Motion and Diffusion Physics, Phase-Contrast MRI).

3. BB Pulse and MSDE Mode

Choosing MSDE as the black-blood pulse gives direction-independent suppression: it acts on the motion of spins along the sensitised axes, not on their position relative to a slice, which is why it is used for intracranial arteries running in every direction within a 3D volume (Vessel Wall Imaging — Intracranial Arteries, Vessel Wall Imaging — Universal Technique Master). The manufacturer's black-blood application page describes the 3D brain use as higher and isotropic resolution compared with its 2D double-inversion methods at the same coverage and scan time, and a reduction of intraluminal blood signal compared with a 3D T1-weighted scan without MSDE pre-pulse; the page gives no parameters and does not define MSDE [11]. These are manufacturer claims, not independent evidence.

With MSDE mode = user defined the remaining fields are set by the operator. No source documents what a preset mode would set, so no statement is made here about presets. Reasoning: since the sub-fields below interact (Section 8), changing one of them from a protocol that was tuned as a set should be done deliberately and one field at a time.

4. Refocusing Pulses (1 or 2)

The conventional MSDE uses one 180° pulse between the two 90° pulses and is highly sensitive to B1 inhomogeneity, with marked signal loss at high m1; iMSDE adds a second 180° pulse, giving four RF pulses and, in the design of Wang and colleagues, four gradient lobes of alternating polarity [8, 9]. In their simulations iMSDE kept higher signal than MSDE over almost all combinations of relative B1 and B0 offset, and in phantom and in vivo at high m1 MSDE lost considerable signal while iMSDE retained it [9]. In the patent's in vivo test (5 volunteers) iMSDE gave a higher sternocleidomastoid SNR (values 11.4 ± 4.7 and 16.7 ± 7.7, p < 0.001) and a slightly but significantly lower lumen SNR (3.1 ± 1.0 and 2.7 ± 0.9, p = 0.001) [8]; the order of the values in the retrieved summary was not matched to the two sequences, so only the direction is stated.

A later comparison for 3D imaging (Balu and colleagues, 2011; fixed gradient strength 30 mT/m) found at m1 = 100, 50 and 25 mT·ms²/m an SNR of 8.01, 8.33 and 8.91 for iMSDE against 5.46, 6.09 and 6.27 for the single-refocusing bipolar MSDE, and inhomogeneity indices (ideal 1.0) of 0.83, 0.67 and 0.83 against 2.50, 2.00 and 2.50. Lowering the gradient strength from 30 to 10 mT/m raised the single-refocusing SNR from 5.46 to 7.92 and brought its inhomogeneity to 1.00, which the authors attribute to eddy currents; iMSDE stayed near 8–9 [7]. The authors conclude that iMSDE is preferable at moderate and low m1 and that the single-refocusing version suits only low m1 with low gradient strength [7].

Use. Reasoning from [7, 9]: two refocusing pulses is the choice with the stronger support whenever the m1 is not very low, including at 3 T where B1 inhomogeneity is relevant; one refocusing pulse is justified only when a short preparation matters more than uniformity and the sensitivity is low. The cost of the second pulse is a longer preparation and more RF per repetition (Sections 6 and 8). Cervantes and colleagues used two refocusing pulses on each axis to reduce sensitivity to transmit-field effects [4]. The abstracts above compare designs; none is a clinical accuracy study.

5. Refocusing Type (Adiabatic)

The refocusing pulses may be composite pulses, as in the original iMSDE (composite 90y–180x–90y) [9], or adiabatic pulses, which are less sensitive to B1 inhomogeneity but longer. The iMSDE patent states that adiabatic pulses could be used and would substantially lengthen the RF pulses and therefore TEprep [8]. A GE abstract on a slice-selective MSDE module states that its refocusing pulses are adiabatic and that this reduces B1 sensitivity, especially at 3 T [6]. Wang and colleagues (Philips Research and University of Washington) built an "inhomogeneity-insensitive" MSDE (i2MSDE) by inserting the motion-sensitising gradients into a B1-insensitive rotation (BIR-4) adiabatic pulse; in a phantom and one volunteer, the banding seen with iMSDE at the periphery of the field of view, at 30, 75 and 90 mm off-centre in the volunteer, was removed, with no difference in lumen SNR, at matched m1 of about 1500 mT·ms²/m [5].

Use. Reasoning from [5, 6, 8]: an adiabatic setting is the option aimed at B1-related signal drop and banding in large-coverage 3D acquisitions at 3 T, particularly away from the isocentre; its cost is a longer preparation (Figure 1) and presumably more RF energy, which was not quantified in these sources. Whether the Philips "adiabatic" option is the BIR-4-based i2MSDE of [5] or another adiabatic refocusing design was not established, and the sources describe i2MSDE without stating its console name.

6. TEprep

TEprep is the total time between the first and last 90° pulses; within it the intervals are TEprep/4, TEprep/2 and TEprep/4 [8]. Two things depend on it. First, the first moment: m1 is set by gradient strength and duration, and the duration also sets TEprep, so a longer preparation allows a larger m1 [8]. Second, T2 decay: signal of stationary tissue is attenuated during the preparation, and the patent identifies the T2 decay in a longer TEprep as a limiting factor, increasing as stronger, longer gradients are used to suppress very slow flow [8]. For this reason the patent suggests starting from the strongest available gradient (to keep TEprep short), sweeping gradient durations from about 0 to 10 ms in 1 ms steps, and choosing the setting that maximises wall-to-lumen CNR [8].

Figure 1 — Signal remaining after the preparation as a function of TEprep

Reported preparation durations span a wide range: 8 ms in the iMSDE simulations, 19.4 ms (MSDE) and 21.2 ms (iMSDE) in the phantom [8], 8.0–9.8 ms (iMSDE) in the 3D comparison of Balu and colleagues at m1 = 25–100 [7] and 57 ms for peripheral-nerve imaging at m1 of 3915 mT·ms²/m and above [4]. A TEprep of 50 ms therefore lies in the range reported for very high sensitivity rather than for the shortest preparations. In Figure 1 (illustrative T2 values; the 200 ms value is the in vivo value of the patent simulations) the fraction of signal remaining at 50 ms is 0.78, 0.61 and 0.37 for T2 of 200, 100 and 50 ms, against 0.90, 0.81 and 0.65 at 21.2 ms. Reasoning: the T2 loss applies to the vessel wall as well as to blood, so TEprep trades wall SNR against the m1 it permits; and since an adiabatic refocusing pulse lengthens the preparation [8], the choice in Section 5 and the value here are linked.

Use. What the console does when TEprep is set to a value shorter than the gradients and pulses require, and what it computes in the non-user mode, were not documented in any source found. Reasoning: given that the achievable m1 grows with the time available [8], the safest interpretation is to treat TEprep and Venc as a pair (Section 8).

7. Venc (RL, AP, FH)

7.1 Relation to the First Moment

The cited MSDE sources describe the sensitivity by m1, not by VENC; the patent does not mention VENC [8]. If VENC is defined as the velocity that produces a phase of π, then VENC = π / (γ · m1); in practical units VENC (cm/s) ≈ 1174 / m1 (m1 in mT·ms²/m). This is a calculation on this page. Applied to the m1 values reported in the sources it gives:

Figure 2 — VENC as a function of the first gradient moment, with the m1 values of the cited sources

A console value of 0.5 would correspond to m1 ≈ 2350 mT·ms²/m if the field is in cm/s and uses this definition — neither was verified. On that assumption it lies between the highest values of the intracranial and carotid studies (about 1500–1581) and the nerve values (3915 and above). The page Vessel Wall Imaging — Intracranial Arteries states, in its Section 10.1, that the velocity sensitivity for intracranial vessel wall imaging is "typically 2–4 cm/s"; a source for that range was not checked in this session, and 2–4 cm/s would correspond to m1 of roughly 290–590, near the value of Wang 2007 [10]. The discrepancy with the console value is flagged for confirmation, and no correction has been made to that page.

7.2 What a Given VENC Suppresses

Figure 3 — Longitudinal signal restored by the tip-up as a function of velocity in units of VENC

In the ideal model the signal returned to the longitudinal axis is M0·cos(π v / VENC) for a single velocity component v along the sensitised direction (Figure 3). Four consequences follow; they are derived on this page and are not findings of the cited sources. (1) Stationary tissue (v = 0) is kept. (2) The first null is at v = VENC/2, not at VENC: blood moving more slowly than about VENC/2 is only partially suppressed, so a smaller VENC (larger m1) is what reaches slow flow, such as in aneurysm sacs or recirculation zones, and slow CSF motion. (3) The model signal returns to full magnitude, with inverted sign, at v = VENC, so suppression is not monotonic with velocity for a single velocity; in a voxel containing a spread of velocities the average smooths the null and reduces the revival (dashed curves). (4) Moving tissue near the wall is also sensitised, which is the reasoning behind the reduction of wall signal close to the lumen noted on Vessel Wall Imaging — Intracranial Arteries when sensitivity is set too high; with a smaller VENC this risk grows. Zhou and colleagues observed the practical form of the trade-off: larger m1 suppressed CSF better, but the longer preparation reduced vessel wall SNR [3]. The magnitude of SNR loss reported for MSDE in other studies is quoted as 20–40% by Xie and colleagues, as a secondary statement [2].

7.3 The Three Axes (RL, AP, FH)

The sources describe gradients along the X, Y and Z axes [8] and application along three orthogonal directions [3]; none states whether the three axes are played together in one preparation or in separate acquisitions. Reasoning: if the three sensitising gradients act simultaneously, the effective sensitisation is a single vector whose components are the per-axis m1 values; the phase is the dot product of that vector with the velocity, so flow perpendicular to the effective direction acquires no phase and escapes suppression, and equal values on the three axes make this direction oblique to the anatomy. If the axes are applied separately, each direction is covered independently. Which applies to this console was not determined; this is one of the items in Section 9.

8. Interaction of the Fields and Reading the Listed Set

• Venc ↔ TEprep. A stronger sensitivity (smaller Venc, larger m1) needs more gradient area and therefore more time [8]; the cost is T2 loss (Figure 1) and, in practice, lower wall SNR [3].
• Refocusing pulses ↔ TEprep. The second 180° and adiabatic pulses add duration [8].
• Refocusing type ↔ coverage and field strength. The adiabatic option addresses B1-related signal drop and banding, mainly in large-volume 3D at 3 T [5, 6].
• Preparation ↔ readout. A reduction of wall SNR and wall CNR with MSDE is a recurring finding: in 14 healthy volunteers on Philips 3 T, wall SNR and wall-to-lumen and wall-to-CSF CNR were higher without the black-blood preparation (SPACE with and without DANTE) than with the Philips 3D TSE with and without iMSDE, while lumen SNR was higher with the Philips sequence, and visual blood-suppression scores did not differ significantly among the four sequences [1]. The comparison used different scanners and coils for the two vendors, and the authors consider TR and ETL not fully optimised [1].

The listed values read together. Reasoning: two adiabatic refocusing pulses, TEprep 50 ms and Venc 0.5 on all three axes describe a preparation designed for strong sensitisation (if the unit is cm/s, about 2350 mT·ms²/m, Section 7.1), robust to B1 error through the number and type of refocusing pulses, and paid for by a long preparation with the signal attenuation shown in Figure 1. Whether that set suits a given protocol cannot be judged from the field values alone: it depends on the readout (TR, ETL, voxel size), the clinical target (blood, slow venous flow, CSF) and the accepted wall-SNR cost.

8.1 Troubleshooting Guide Derived From the Above

9. Evidence, Limits and Items to Confirm on the Console

• No guideline or consensus statement addresses these parameters, and no vendor document defining the console fields was found.
• The evidence is technical: one patent, six conference abstracts, two journal studies in volunteers or patients and one secondary statement. Most were read at abstract level or from retrievable text; no study read measures diagnostic accuracy as a function of these settings.
• The VENC–m1 conversion, the signal-versus-velocity model and the T2 attenuation are calculations with stated assumptions (single velocity or uniform spread, perfect refocusing, illustrative T2). Real flow is laminar and pulsatile.
• The original journal descriptions of MSDE and of iMSDE by Wang and colleagues were not read (the PubMed and PubMed Central pages were blocked in this session; the iMSDE paper appears in search results under the title "Enhanced image quality in black-blood MRI using the improved motion-sensitized driven-equilibrium (iMSDE) sequence" but was not opened), so those facts are cited from conference abstracts and the patent.

To be confirmed on the console or in the Philips documentation: (1) the options of BB pulse, MSDE mode and refocusing type and what the non-user mode sets; (2) the unit of Venc and its definition; (3) whether the three axes are applied simultaneously; (4) what TEprep does when it is shorter than feasible and what it is in automatic mode; (5) the value of MTC, which is outside the scope of this page.

10. Evidence-Based References

No guideline or consensus statement was identified for MSDE parameters. References are listed by tier and recency; conference abstracts and the patent were read in the retrievable text only.

C. Comparative and Original Studies

D. Technical Abstracts and Patent

End of document — Philips MSDE Black-Blood Preparation — MRIninja v1.0 — October 2026