NMR and MRI physics · Introductory

T2 decay (transverse)

Written by SignalsLab

Early edition: these lessons are being reviewed.

Content

Summary

T2 decay is the loss of transverse magnetization after an RF pulse, as the spins that were precessing together drift out of phase. The signal falls exponentially with time constant T2, and this decay sets the contrast in T2-weighted MRI images.

Intuition

Imagine runners starting a lap together at the same instant, each running at a slightly different speed. At first they are bunched, and the crowd as a whole is easy to see. As time passes they spread around the track, and the group's net position averages toward nothing. Spins behave the same way. After a 90 degree pulse all the proton magnetic moments point together in the transverse plane. Each one feels a slightly different local magnetic field from neighbouring molecules, so each precesses at a slightly different frequency. The vector sum shrinks even though no energy needs to be lost from each spin. Tissues where water moves freely, such as cerebrospinal fluid, have long T2 and stay bright. Tissues with tightly bound water, such as tendon, have very short T2 and fade quickly. Real images also suffer from static field inhomogeneity, which makes decay faster, called T2*. A spin echo refocuses that part and recovers true T2.

Key equations

Mxy(t)=M0e−t/T2
(1)

Transverse magnetization decays exponentially with time constant T2, starting from M0 right after a 90 degree pulse.

dMxydt=−MxyT2
(2)

The Bloch equation for the transverse component: the decay rate is proportional to the current magnetization.

1T2∗=1T2+γΔB0
(3)

Observed decay rate adds a contribution from field inhomogeneity (with γ the gyromagnetic ratio and ΔB0 the field spread).

S(TE)∝e−TE/T2
(4)

Signal at echo time TE in a spin echo sequence determines T2 weighting.

Derivation

  1. Start with the Bloch equation for transverse magnetization in the rotating frame, ignoring off-resonance precession: the rate of change is proportional to minus the magnetization divided by T2.
  2. Write this as dMdt=−MT2.
  3. Separate variables: dM/M = -dt/T2.
  4. Integrate from time 0 to t: ln(M(t)) - ln(M(0)) = -t/T2.
  5. Exponentiate: M(t) = M(0) e^(-t/T2).
  6. Right after a 90∘ pulse M(0) equals the equilibrium value M0, giving the final result.
  7. At t=T2 the magnetization equals M0/e, about 37 percent of its starting value.

Worked example

Problem

A tissue has T2 = 80 ms. A spin echo sequence uses TE = 40 ms. What fraction of the transverse magnetization remains at the echo, and how long until it falls to 5 percent?

  1. Fraction at TE: e^(-40/80) = e^(-0.5).
  2. e^(-0.5) is approximately 0.607.
  3. For 5 percent: e^(-t/80) = 0.05.
  4. Take logs: -t/80 = ln(0.05) = -3.00.
  5. So t = 3.00 x 80 = 240 ms.

Answer About 61 percent remains at TE = 40 ms, and it reaches 5 percent at about 240 ms (roughly 3 T2).

Common mistakes

  • Confusing T2 with T1: T2 is loss of transverse phase coherence, not recovery of longitudinal magnetization, and T2 is always less than or equal to T1.
  • Thinking T2 decay means the spins lose energy to the lattice; it is mainly dephasing through spin-spin interaction.
  • Mixing up T2 and T2*: T2* includes static field inhomogeneity and is shorter; only a spin echo recovers that part.
  • Believing the magnetization reaches zero at t = T2; it has only fallen to about 37 percent.
  • Forgetting that a long TE gives T2 weighting and a short TE reduces T2 contrast.

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(11 lines)
from manim import *

class T2Decay(Scene):
    """Teaching scene: transverse magnetisation decays with T2 (S = S0 e^{-t/T2}). Test scene for the render container."""
    def construct(self):
        title = Text("T2 relaxation", font_size=40).to_edge(UP)
        axes = Axes(x_range=[0, 5, 1], y_range=[0, 1.1, 0.5], x_length=7, y_length=3.5, axis_config={"include_tip": False})
        labels = axes.get_axis_labels(MathTex("t"), MathTex("S(t)"))
        curve = axes.plot(lambda t: np.exp(-t / 1.5), x_range=[0, 5], color=YELLOW)
        eq = MathTex(r"S(t)=S_0\,e^{-t/T_2}").next_to(axes, DOWN)
        self.play(Write(title)); self.play(Create(axes), Write(labels)); self.play(Create(curve), run_time=2); self.play(Write(eq)); self.wait(1)

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.