NMR and MRI physics · Intermediate

Inversion recovery and null point

Written by SignalsLab

Early edition: these lessons are being reviewed.

Content

Summary

Inversion recovery starts with a 180-degree pulse that flips longitudinal magnetization to −M0. It then recovers toward +M0 with time constant T1. After a delay TI, a 90-degree pulse reads the signal. The signal passes through zero at the null point, TI=T1ln⁡2, which can be used to suppress a chosen tissue such as fat or CSF.

Intuition

Picture the magnetization as a vector along the z axis. The 180-degree pulse flips it to point straight down. Spin-lattice relaxation then pulls it back up along an exponential curve. On the way up it must cross zero. A 90-degree pulse tips only the longitudinal component into the transverse plane, so if you excite exactly when a tissue is crossing zero, that tissue gives no signal. Tissues with different T1 cross zero at different times, so the choice of TI selects which tissue disappears. A short T1 crosses early and a long T1 crosses late.

Key equations

Mz(t)=M0(1−2e−t/T1)
(1)

Longitudinal recovery after an ideal 180-degree inversion, starting at −M0.

TInull=T1ln⁡2≈0.693T1
(2)

Inversion time at which Mz crosses zero, so that tissue gives no signal.

S∝|M0(1−2e−TI/T1)|
(3)

Magnitude-reconstructed signal; the sign is lost, giving a V-shaped curve.

Mz(t)=M0(1−(1+cos⁡α)e−t/T1)
(4)

General case for an inversion pulse of flip angle α; α=180∘ gives the ideal formula.

Derivation

  1. Start from the Bloch equation for the longitudinal component: dMzdt=M0−MzT1.
  2. Mz(t)=M0+(Mz(0)−M0)e−t/T1.
  3. An ideal 180-degree pulse sets Mz(0)=−M0.
  4. Substitute: Mz(t)=M0−2M0e−t/T1=M0(1−2e−t/T1).
  5. Set Mz = 0: 1 = 2e^{-TI/T1}, so e^{-TI/T1} = 1/2.
  6. Take logarithms: TI/T1 = ln 2, so TI_null = T1 ln 2.

Worked example

Problem

Fat at 1.5 T has T1 of about 260 ms. Which TI nulls fat in a STIR sequence? At that TI, what fraction of M0 does a tissue with T1 = 900 ms have?

  1. TI = T1 ln 2 = 260 × 0.693 = 180 ms.
  2. For the tissue with T1 = 900 ms: TI/T1 = 180/900 = 0.2.
  3. e^{-0.2} = 0.819.
  4. Mz/M0=1−2(0.819)=1−1.638=−0.638.
  5. The magnitude of the signal is 0.64 M0.

Answer TI is about 180 ms. The 900 ms tissue has Mz=−0.64M0, so its magnitude signal is 0.64M0 while fat is suppressed.

Common mistakes

  • Using TI=T1 instead of T1ln⁡2; at TI=T1 the magnetization is still at −0.26M0.
  • Forgetting that magnitude images discard the sign, so the curve is V-shaped and the negative values look bright.
  • Assuming the null depends on T2; it depends only on T1 (and on repetition time if TR is short).
  • Ignoring incomplete recovery: with short TR the null becomes TI = T1 ln(2/(1+e^{-TR/T1})).
  • Believing STIR suppresses fat chemically; it suppresses any tissue with the matching T1, including some contrast-enhanced tissue.

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(112 lines)
import sys
import numpy as np
from manim import *
from blocks import *

# Inversion Recovery Model: Mz(t) = M0 * (1 - 2*exp(-t/T1))
# Null point occurs when 1 - 2*exp(-t/T1) = 0 => t = T1 * ln(2)

T1_VAL = 1.0  # Normalized T1
M0_VAL = 1.0

def mz(t, t1=T1_VAL, m0=M0_VAL):
    """Longitudinal magnetisation after 180 inversion pulse."""
    return m0 * (1.0 - 2.0 * np.exp(-t / t1))

def null_time(t1=T1_VAL):
    """Calculate the null point time."""
    return t1 * np.log(2)

class Main(Scene):
    def construct(self):
        self.camera.background_color = "#101418"
        
        # Title
        TitleCard(self, "Inversion Recovery", "Mz(t) = M0(1 - 2e^{-t/T1})")

        # Compute curve data
        t_max = 5.0 * T1_VAL
        ts = np.linspace(0, t_max, 500)
        zs = mz(ts)
        
        # Null point
        t_null = null_time()
        z_null = mz(t_null) # Should be 0

        # Plot setup
        # X: 0 to 5*T1, Y: -1.2 to 1.2
        plot = CurvePlot(
            func=lambda t: mz(t),
            x_range=(0, 5.5, 1),
            y_range=(-1.2, 1.2, 0.5),
            x_label="t / T1",
            y_label="Mz / M0",
            x_length=8.0,
            y_length=4.5,
            color=C_MAIN
        )
        plot.shift(LEFT * 1.5 + DOWN * 0.2)
        
        # Draw axes and curve
        self.play(Create(plot.axes), FadeIn(plot.labels))
        self.play(Create(plot.curve), run_time=2.5)

        # Equation
        eq = MathTex(
            r"M_z(t) = M_0 \left( 1 - 2e^{-t/T_1} \right)",
            font_size=30
        )
        eq.to_edge(RIGHT, buff=0.5).shift(UP * 1.5)
        self.play(Write(eq), run_time=1)

        # Highlight Null Point
        # 1. Vertical dashed line from axis to curve
        null_dot = plot.dot_at(t_null, color=C_BAD)
        null_line = DashedLine(
            plot.axes.c2p(t_null, 0),
            plot.axes.c2p(t_null, z_null),
            color=C_BAD,
            stroke_width=2
        )
        
        # 2. Label for Null Point
        null_lab = Text(f"Null Point: t = T1 ln(2)", font_size=26, color=C_BAD)
        null_lab.next_to(null_dot, RIGHT, buff=0.3)
        
        # Animate the null point appearance
        self.play(
            Create(null_line),
            FadeIn(null_dot),
            FadeIn(null_lab),
            run_time=1.5
        )
        
        # Highlight the zero crossing specifically on the axis
        zero_mark = Dot(plot.axes.c2p(t_null, 0), color=C_ACCENT, radius=0.1)
        self.play(FadeIn(zero_mark))

        StepCaption(self, "At t = T1 ln(2) ≈ 0.693 T1, longitudinal magnetisation is zero.")
        self.wait(2)


if __name__ == "__main__" and "--selftest" in sys.argv:
    # Test 1: Initial value is -M0
    assert abs(mz(0.0) - (-1.0)) < 1e-9
    
    # Test 2: Asymptote is +M0
    assert abs(mz(100.0) - 1.0) < 1e-9
    
    # Test 3: Null point calculation
    t_n = null_time()
    assert abs(mz(t_n)) < 1e-9
    
    # Test 4: Null point time value
    expected_t_null = np.log(2)
    assert abs(t_n - expected_t_null) < 1e-9
    
    # Test 5: Curve is monotonic increasing
    ts_test = np.linspace(0, 5, 100)
    zs_test = mz(ts_test)
    assert np.all(np.diff(zs_test) > 0)
    
    print("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.