Higher Resolution on a Twitching Wrist
Ms. L, 41, is referred for MRI of the right wrist after a fall, with ulnar-sided pain and a suspected TFCC tear. Your site is running the 3 T scanner today, and an 8-channel dedicated wrist coil is available. Her removable fiberglass splint comes off in the room. The coronal FSE protocol uses an 80 mm FOV, 256 × 224 matrix, 2.0 mm slices with a 0.4 mm gap, ETL 8, 2 averages, and 200 Hz/px receive bandwidth; the sequence runs a little over three and a half minutes. On the first coronal series there is mild blurring and a faint ghost across the image; the patient admits her hand "twitched" twice during the scan. The radiologist asks you to repeat it at "0.2 mm in-plane resolution."
1. Calculate the current in-plane resolution and voxel volume, then state what matrix would be needed for 0.2 mm in-plane at this FOV and what that does to SNR.
Read the worked answer
Current:
- Frequency direction: 80 mm / 256 = 0.31 mm.
- Phase direction: 80 mm / 224 = 0.36 mm.
- Voxel volume: 0.31 × 0.36 × 2.0 ≈ 0.22 mm³.
Proposed 0.2 mm in-plane at the same 80 mm FOV:
- 80 mm / 0.2 mm = 400 phase-encoding steps in each in-plane direction (a 400 × 400 matrix).
- Voxel volume: 0.2 × 0.2 × 2.0 = 0.08 mm³, about 36% of the current volume.
Consequence: SNR scales with voxel volume, so signal falls to roughly one-third. To restore the original SNR you would need about (1/0.36)² ≈ 7.8× the averages, i.e., a scan approaching 30 minutes for a single series - unacceptable for a wrist and a moving patient. The honest answer to the radiologist is that 0.2 mm in-plane is achievable but only by trading time and motion robustness, and at 3 T the practical wrist limit sits near 0.25–0.3 mm in-plane with a dedicated coil.
2. Given that calculation, what would you actually change to improve diagnostic yield for a TFCC question, and why?
Read the worked answer
Prioritize what the surgeon needs: thin slices and high in-plane resolution through the triangular fibrocartilage, with enough SNR to see a tear.
- Slice thickness before in-plane matrix. Going from 2.0 mm to 1.5 mm slices costs 25% of the signal but improves through-plane partial volume at the thin TFCC; combining 1.5 mm slices with a 320–352 in-plane matrix is a better trade than 0.2 mm in-plane at 2 mm slices.
- Recover SNR elsewhere: keep the smallest FOV that still covers the anatomy, use the wrist coil with the element closest to the ulnar side, lower receive bandwidth if TE permits, and use 3–4 averages rather than 2.
- Buy back time: apply parallel imaging if the coil geometry supports it, shorten ETL appropriately, and drop a non-contributing sequence rather than adding series.
- Consider a 3D sequence for thin, contiguous slices, remembering that 3D multiplies scan time by the number of partitions and that SNR per partition is lower.
The principle: resolution, SNR, and time form a fixed budget. Moving one dial requires an explicit decision about which of the other two you are spending.
3. The images are blurred with a faint ghost. How do you decide whether this is motion or something else, and what is your fix?
Read the worked answer
Is it motion?
- Ghosting that repeats across the phase-encoding direction and involves only some slices or a single series points to motion; the patient's history supports it.
- Blurring that affects all series, all coils, and a phantom is a hardware or shim problem, not motion.
- Check whether the blurring is directional (phase direction worse) and whether the ghost spacing is periodic - a phase-encode ghost from pulsation would be periodic and reproducible, which is unlikely in the wrist.
Fix: More averages will improve SNR but will only partially suppress ghosting and will lengthen an already long scan in a patient who is already moving. The higher-yield moves are:
- Immobilize properly: sandbag, strap the forearm and hand, and place the wrist in neutral; confirm the coil is snug but not compressing.
- Coach between series: short breath-hold-like instruction, "stay still until I tell you," and verbal feedback over the intercom.
- Shorten the sequence (parallel imaging, modest ETL increase) so the acquisition fits inside a tolerable still period.
- If available, use a motion-robust option such as a radial or blade-type FSE readout.
Then repeat the single affected series rather than the whole protocol.