NMR and MRI physics · Advanced

Spoiled gradient echo and the Ernst angle at steady state

Written by SignalsLab

Early edition: these lessons are being reviewed.

Content

Summary

A spoiled gradient echo (SPGR/FLASH) sequence repeats a low flip angle pulse every TR and destroys residual transverse magnetization each cycle, so only longitudinal recovery carries forward. The signal reaches a steady state that depends on T1, TR and flip angle, and is maximal at the Ernst angle, cos(alpha_E) = exp(-TR/T1).

Intuition

Each RF pulse tips some longitudinal magnetization into the transverse plane, where it gives signal; the rest stays along z. If TR is short compared with T1, z magnetization has little time to recover, so a large flip angle quickly saturates the spins and the signal collapses after a few pulses. A small flip angle uses only a fraction of Mz each time, leaving more for the next pulse, but yields less signal per pulse. The best compromise is the Ernst angle: small enough to preserve Mz, large enough to give transverse signal. Spoiling (gradient dephasing plus RF phase cycling) ensures no transverse coherence builds up, so the simple T1-only model holds. Short TR relative to T1 means a small Ernst angle; as TR approaches several T1, the Ernst angle approaches 90 degrees.

Key equations

S=M0sin⁡α(1−E1)1−E1cos⁡αe−TE/T2∗,E1=e−TR/T1
(1)

Steady-state spoiled gradient echo signal

cos⁡αE=e−TR/T1
(2)

Ernst angle that maximizes the signal

Mz+=Mz−cos⁡α,Mz−(n+1)=M0(1−E1)+Mz+(n)E1
(3)

Effect of pulse and recovery over one TR

αE≈2TR/T1(TR≪T1)
(4)

Small-TR approximation, in radians

Derivation

  1. Assume perfect spoiling: transverse magnetization is zero just before each pulse.
  2. Just before a pulse, longitudinal magnetization is Mz−. The pulse leaves Mz+=Mz−cos⁡(α) and creates transverse magnetization Mz−sin⁡(α).
  3. During TR, Mz recovers: Mz(next)=M0(1−E1)+Mz+E1, with E1=exp⁡(−TR/T1).
  4. At steady state Mz+=Mz−, so Mz−=M0(1−E1)/(1−E1cos⁡α).
  5. Signal is proportional to Mz−sin⁡(α)×T2∗ decay at TE, giving the signal equation.
  6. Set dS/dα=0: cos⁡(α)(1−E1cos⁡α)−sin⁡(α)(E1sin⁡α)=cos⁡(α)−E1=0, so cos⁡(αE)=E1.

Worked example

Problem

Brain tissue with T1 = 1000 ms is imaged with spoiled GRE at TR = 20 ms. Find the Ernst angle and the relative signal at the Ernst angle versus at 90 degrees.

  1. E1 = exp(-20/1000) = 0.9802.
  2. cos(alpha_E) = 0.9802, so alpha_E = arccos(0.9802) ≈ 11.4 degrees.
  3. Signal factor at Ernst: sin⁡(11.4∘)(1−0.9802)/(1−0.9802×0.9802)=0.1977×0.0198/(1−0.9608)=0.003914/0.0392≈0.0999 (times M0).
  4. At 90 degrees: 1×(0.0198)/(1−0)=0.0198 (times M0).
  5. Ratio ≈ 0.0999/0.0198 ≈ 5.

Answer Ernst angle ≈ 11°; its signal is about five times that at 90 degrees (≈0.10 M0 vs 0.02 M0).

Common mistakes

  • Using the Ernst angle formula without spoiling; in unspoiled steady state (e.g. SSFP) T2 matters and the formula fails.
  • Thinking the Ernst angle maximizes contrast; it maximizes signal of one T1, while T1 weighting needs a larger angle.
  • Forgetting that the Ernst angle depends on tissue T1, so it is optimal for only one tissue.
  • Mixing degrees and radians in the small-angle approximation.
  • Ignoring imperfect spoiling or B1 inhomogeneity, which make the actual flip angle differ from nominal.

Animation

Animation coming soon

The animated version of this lesson is being prepared for the web. In the Signals AI app, lessons are drawn step by step.

Code

scene.py(100 lines)
from manim import *
import numpy as np
import sys
from blocks import *
from mri_blocks import *

# Parameters (ms)
T1, T2S, TR, TE = 500.0, 30.0, 10.0, 4.0
ALPHA = np.deg2rad(30)
N_PULSES = 40


def simulate_spoiled(alpha, n=N_PULSES):
    """Run n TRs with ideal spoiling; return lists of Mz before each pulse and signal at TE."""
    M = np.array([0.0, 0.0, 1.0])
    mz_before, sig = [], []
    for _ in range(n):
        mz_before.append(M[2])
        events = [{"type": "rf", "flip": alpha}, {"type": "wait", "dt": TE},
                  {"type": "wait", "dt": TR - TE}]
        _, hist = bloch_simulate(events, M, T1, T2S, [0.0])
        sig.append(float(np.hypot(*hist[0, 2, :2])))   # |Mxy| at TE
        M = hist[0, -1].copy()
        M[0] = M[1] = 0.0                               # spoiler kills transverse M
    return np.array(mz_before), np.array(sig)


def ss_signal(alpha_rad):
    """Analytic spoiled GRE steady state, including TE decay."""
    E1 = np.exp(-TR / T1)
    return np.sin(alpha_rad) * (1 - E1) / (1 - E1 * np.cos(alpha_rad)) * np.exp(-TE / T2S)


def ernst_angle():
    """Ernst angle in radians."""
    return float(np.arccos(np.exp(-TR / T1)))


class Main(Scene):
    def construct(self):
        TitleCard(self, "Spoiled GRE: steady state", "Ernst angle maximises signal")
        mz, sig = simulate_spoiled(ALPHA)

        # Timing diagram for one TR
        seq = PulseSequenceDiagram(rows=("RF", "Gz", "Gx", "ADC"), t_max=TR, width=11, height=2.7)
        seq.move_to(UP * 1.35)
        parts = VGroup(
            seq.add_rf(0.0, 1.2, kind="sinc", label="30°"),
            seq.add_gradient("Gz", 0.0, 1.2, amp=0.8),
            seq.add_gradient("Gz", 7.0, 9.6, amp=1.0, color=C_BAD),
            seq.add_gradient("Gx", 1.5, 2.6, amp=-0.8),
            seq.add_gradient("Gx", 2.8, 5.2, amp=0.8),
            seq.add_adc(2.8, 5.2),
            seq.add_marker(TE, "TE"),
            seq.add_marker(TR, "TR"),
        )
        self.play(FadeIn(seq), FadeIn(parts), run_time=1.5)
        cursor = seq.time_cursor()
        self.add(cursor)

        # Lower left: signal per pulse
        left = SignalPlot(x_range=(0, 40, 10), y_range=(0, 0.5, 0.25), x_label="pulse n",
                          y_label="S", width=5, height=1.7).move_to(LEFT * 3.4 + DOWN * 1.8)
        self.play(FadeIn(left), run_time=0.6)
        StepCaption(self, "Each TR: excite, read at TE, spoil the leftover Mxy")
        dots = VGroup()
        for k in range(6):
            self.play(seq.cursor_to(TR), run_time=0.75, rate_func=linear)
            d = left.dot_at(k + 1, sig[k])
            dots.add(d)
            self.add(d)
            seq.tracker.set_value(0.0)
        rest = VGroup(*[left.dot_at(k + 1, sig[k], color=C_MAIN) for k in range(6, N_PULSES)])
        self.play(FadeIn(rest, lag_ratio=0.1), run_time=1.2)
        StepCaption(self, "Mz saturates: signal settles to a steady state")

        # Lower right: steady-state signal vs flip angle
        right = SignalPlot(x_range=(0, 90, 30), y_range=(0, 0.12, 0.04), x_label="flip °",
                           y_label="Sss", width=5, height=1.7).move_to(RIGHT * 3.4 + DOWN * 1.8)
        a = np.linspace(0, 90, 181)
        curve = right.plot(a, ss_signal(np.deg2rad(a)))
        self.play(FadeIn(right), Create(curve), run_time=1.5)
        ae = np.rad2deg(ernst_angle())
        ed = right.dot_at(ae, ss_signal(ernst_angle()), color=C_GOOD)
        el = MathTex(r"\alpha_E=%.1f^\circ" % ae, font_size=30, color=C_GOOD).next_to(ed, UP, buff=0.1).shift(RIGHT * 0.6)
        sd = right.dot_at(30, sig[-1], color=C_ACCENT)
        self.play(FadeIn(ed), Write(el), FadeIn(sd), run_time=1)
        StepCaption(self, "Takeaway: cos(alpha_E) = exp(-TR/T1) gives maximal spoiled-GRE signal")
        self.wait(1.5)


if __name__ == "__main__" and "--selftest" in sys.argv:
    mz, sig = simulate_spoiled(ALPHA)
    assert abs(mz[0] - 1.0) < 1e-12
    assert abs(sig[0] - np.sin(ALPHA) * np.exp(-TE / T2S)) < 1e-9
    assert abs(sig[-1] - ss_signal(ALPHA)) < 1e-3
    a = np.linspace(0.001, np.pi / 2, 20000)
    assert abs(a[np.argmax(ss_signal(a))] - ernst_angle()) < 1e-3
    assert np.all(np.diff(sig) <= 1e-12)
    print("selftest ok")

The Manim scene code that draws this lesson's animation in the Signals AI app (Python, Manim Community).

ccBYNC© 2026 SignalsLab. Lesson text shared as is under CC BY-NC 4.0.