Spin echo sequence

A 90° pulse, a 180° pulse at half the echo time, and an echo that has forgotten the field's imperfections. Spin echo is why T2-weighted MRI looks clean near air and metal, and fast spin echo is why it no longer takes over ten minutes.

By the ScanFlo Editorial Team · Updated 12 October 2026

A spin echo sequence is an MRI pulse sequence that uses a 90° excitation pulse followed by a 180° refocusing pulse at half the echo time. The 180° pulse reverses the dephasing caused by magnetic field inhomogeneity, so the signal at TE depends on true T2 rather than T2*. That is why spin echo images hold up well near air and metal, and its fast spin echo form is the standard way to make T2-weighted images.

Pulse timing and refocusing: IAEA handbook, §14.5.5 and §15.4 and Radiopaedia, Spin echo sequences. History: Hahn (1950) and Hennig et al. (1986).

What happens inside one spin echo

The 90° pulse tips the magnetisation into the transverse plane. The spins start in phase and immediately begin to fan out. Part of that is true T2 decay, random interactions between neighbouring spins. Part is the field itself: imperfect shimming, and the field distortion at air–tissue boundaries, make some spins precess a little faster than others. Together these give the rapid T2* decay of the free induction decay.

At TE/2 the 180° pulse flips the spins over. Spins that had raced ahead are now behind, and the slow ones are in front. They keep precessing at the same rates, so the fast ones catch up, and at TE they all line up again: the echo. The field-related dephasing cancels exactly because it is static. The T2 part does not, because it was random. The echo is therefore smaller than the starting signal by T2 decay alone (IAEA, §14.5.5).

Worked example: cancelling a field error

At 1.5 T protons precess at about 63 MHz (42.58 MHz per tesla; IAEA Table 14.1). Take a spin next to the sinuses where the local field is 1 part per million high. It precesses about 63 Hz faster than its neighbours. With TE = 100 ms, the 180° pulse comes at 50 ms, by which time that spin is 63 × 0.05 ≈ 3.2 cycles ahead, more than enough to destroy its contribution to the voxel's signal.

The 180° pulse turns "3.2 cycles ahead" into "3.2 cycles behind". Over the next 50 ms the spin gains another 3.2 cycles at the same rate and arrives at TE exactly in phase. A gradient echo with the same TE has no such reversal, which is why it loses signal near the sinuses and metal while the spin echo does not. That is also the limit: the correction only works for errors that stay constant while the echo forms. Motion and flow between the pulses are not undone.

Why spin echo matters at the console

The power deposition limit is noted in the IAEA handbook, §15.4.2. For how TR and TE turn the same sequence into T1, PD or T2 contrast, read TR and TE and T1 vs T2 weighting.

Fast spin echo: the version you actually run

Conventional spin echo collects one line of k-space per TR, so a 256-line T2 scan at TR 3,000 ms takes almost 13 minutes (scan time = TR × lines × averages; UCLA M219 lecture). Fast spin echo keeps going after the first echo: more 180° pulses, each echo phase-encoded for a different line. The IAEA handbook describes 16, 32 or more lines per TR, and single-shot versions that fill the whole of k-space after one excitation. Divide the 13 minutes by an echo train of 16 and you are under a minute.

Two things change. Each line now has its own echo time, so the TE on the console is an effective TE: the echo time of the lines placed at the centre of k-space, which set the contrast. And later echoes are weaker, because T2 decay continues along the train, which costs some sharpness with very long trains (IAEA, §15.4.3).

Common mistakes

• Expecting spin echo to fix motion. The refocusing pulse cancels static field errors only; patient motion and flow between pulses still blur and ghost.
• Reading the console TE on a fast spin echo as the time of every echo. It is the effective TE; the rest of the train spans a range of echo times.
• Swapping spin echo for gradient echo near metal to save time. The gradient echo will show the susceptibility loss that the 180° pulse would have removed.
• Ignoring SAR when lengthening the echo train or adding slices. The scanner may lengthen TR or cut slices to stay within the RF limit.
• Assuming a long echo train is free. Very long trains reduce signal in the late echoes and blur fine detail.