MRS Data Analysis and Quality Control

47 flashcardspar Moll
Réviser ces flashcardsGratuit · répétition espacée · iPhone, Android, Mac et web

Toutes les cartes

1
What are the three stages of MRS data analysis?
1) Preprocessing: everything done to the raw data to prepare it for fitting (combine coils, align and average, correct artefacts, remove water).
2) Fitting: estimate each metabolite's signal (arbitrary units) with a model, usually a linear combination of basis spectra.
3) Quantification: convert those numbers into reportable units (mmol/kg wet weight, or ratios).
Quality control runs through all three.
2

"Garbage in, garbage out": what does it mean for MRS?
A good fitting algorithm cannot rescue bad data: motion, poor shim, poor water suppression, lipid contamination, eddy currents or chemical-shift artefacts all give unreliable concentrations.
Good data fitted with the wrong settings are also unreliable. Fitting software is automatic but not push-button: it must be adapted to the sequence, TE, B0 and nucleus.
3

What does each FID (time-domain) parameter correspond to in the spectrum?
1) Frequency → peak position.
2) Amplitude → peak area.
3) Damping rate (1/T2*) → linewidth (FWHM = damping/π).
4) Phase at t = 0 → zero-order phase.
5) Beginning time (delay before recording) → first-order phase.
4

Zero-order vs first-order phase error
Zero-order: the same phase error on every peak (from the receiver/coil phase); all peaks mix absorption and dispersion alike.
First-order: the error grows linearly with frequency. Cause: recording starts a delay t0 after the pulse, so each component has already turned by 2πf·t0 (faster components further).
Both are corrected in preprocessing or included in the fit.
5

Are SNR and linewidth enough to judge spectral quality?
No. They are necessary but not sufficient: two spectra can have the same SNR and FWHM while one is ruined by residual water, lipid contamination or a spurious echo.
Quality control means looking for artefacts, then correcting or rejecting the data.
6

What sets the SNR of an MRS spectrum?
Sample and protocol: voxel volume (signal ∝ volume), number of averages (√N), sequence (e.g. STEAM keeps half), TE and TR.
Adjustments: RF power (flip angle), coil tuning/matching, B1 shimming, slice rephasing, gradient delays.
Hardware: receiver noise figure, the coil (e.g. a cryoprobe).
Small peaks vanish first as SNR drops.
7

Besides the shim, what broadens lines or distorts the lineshape?
Broadening: hardware performance, subject motion, and averaging shots whose frequency has drifted (a sum of shifted peaks is a wider peak).
Distorted (asymmetric) shape: a misadjusted shim term (e.g. Z2) or eddy currents. A bad shape misleads the fit even if the width looks fine.
8

Poor water suppression: causes and appearance
Causes: type of WS and its settings (RF power, delays), sensitivity to T1 and B1, broad water from a bad shim (spills outside the suppression band), water from outside the voxel.
Appearance: large residual water at 4.7 ppm, distorted baseline and amplitude/phase distortions of nearby peaks.
9

Lipid contamination: causes and remedies
Fat outside the voxel (scalp, muscle) leaks in through imperfect localization or slice profiles, weak crushers, insufficient OVS, CSDE or motion → broad peaks at 0.9–1.3 ppm that swamp lactate, alanine and macromolecules.
Remedies: OVS bands, double localization (OVS + sequence, or LASER), stronger crushers, high-bandwidth pulses.
10

What is the "pseudo-echo" from water outside the VOI?
Water far from the voxel sits in a very different field (strong 2nd-order shims make distant fields far off), so it is off-resonance, escapes water suppression, and may not be fully crushed. It refocuses late as a delayed echo inside the FID.
In the spectrum: a spurious ringing feature. Remedies: stronger crusher gradients, OVS.
11

Minimum quality criteria for an MRS spectrum (consensus)
Signal: SNR > 3 for major resonances; adequate linewidth; symmetric lineshape; good water suppression; no lipid contamination; no other visible artefacts; watch for hidden ones (CSDE, localization errors, signal cancellation).
Fit: acceptable CRLBs; residuals without unexplained features.
12

Why is automatic quality control especially important for clinical MRS?
Preclinical: anaesthetised animal, little motion, time to shim/WS/OVS, MRS experts at the scanner, moderate data.
Clinical: awake patients, motion, little time, often non-expert operators, huge amounts of data. Artefacts slip through, and standard software without QC can report false changes (e.g. apparent Cho/Cr changes in a glioma) → wrong decisions.
13

Why preprocess before fitting?
Fitting models assume a clean, ideal signal and do not account for motion, frequency drift, eddy currents or misaligned shots, so these would bias the fit.
Corrections should be automated and built into the analysis software.
14

Typical single-voxel preprocessing pipeline, in order
1) Combine RF coil channels.
2) Remove motion-corrupted shots.
3) Correct frequency and phase drift (align shots).
4) Combine (average) shots.
5) Eddy-current correction.
6) Remove residual water (HSVD).
7) Phase correction, then fit.
(Zero-filling and apodization: display only.)
15

Why align frequency and phase before averaging?
B0 drifts over a long scan and motion changes phase. Averaging misaligned shots broadens the peaks and loses signal.
For add/subtract methods (ISIS, SPECIAL), misalignment also leaves signal from outside the voxel after subtraction.
16

How is eddy-current distortion corrected?
Acquire the unsuppressed water signal from the same voxel with the same sequence. Its time-varying phase reflects the eddy-current field; subtract that phase, point by point, from the metabolite FID → clean lineshape.
17

Zero-filling and apodization: when should you use them?
For display only, not before fitting.
Zero-filling pads the FID with zeros: a finer frequency grid, no new information. Apodization (line broadening) multiplies the FID by a decaying window: less noise, broader peaks. Same trade-offs as in k-space for imaging.
18

Why is fitting high-field, short-TE 1H spectra difficult?
Complexity: many resonances, heavy overlap, macromolecules and residual water underneath.
Unpredictable lineshape: residual eddy currents, field inhomogeneity.
Unpredictable baseline: macromolecules, lipids, partly suppressed water.
19

Peak fitting vs HLSVD vs basis-set fitting
Peak fitting: fit each peak with a Lorentzian/Gaussian/Voigt shape; fine for few well-separated peaks (X-nuclei), fails with overlap.
HLSVD: splits the FID into damped sinusoids with no chemical prior knowledge; fast, used for water removal and to quantify simple spectra (jMRUI).
Basis-set (linear combination): spectrum = Σ c_m × basis_m + macromolecules + baseline; the standard for overlapped 1H spectra.
20

What goes into a basis set?
One spectrum per metabolite for exactly the sequence, TE and B0 used, either measured (metabolite solutions) or simulated quantum-mechanically from known chemical shifts and J-couplings (NMRScopeB, Vespa, GAMMA, GAVA).
Plus a macromolecule component (measured with metabolite-nulling inversion, since MM T1 is short, or modelled) and lipids where relevant (e.g. tumours).
21

Why are some metabolites reported as sums (tCr, tCho, Glx)?
When their patterns overlap too much to separate reliably, the fit reports the combination: NAA + NAAG, Glu + Gln (Glx), GPC + PCho (tCho), Cr + PCr (tCr).
22

Common MRS fitting software and how they differ
Vespa, TARQUIN and jMRUI are free; LCModel is listed as paid on the slide. All accept all data types and include preprocessing.
Lineshape: TARQUIN uses Voigt, jMRUI Lorentzian, LCModel estimates it from the data. LCModel needs a supplied basis set and models the baseline with splines.
23

What is the Cramér–Rao lower bound (CRLB)?
The theoretical minimum uncertainty (standard deviation) of a fitted amplitude, given the noise and the model, usually quoted as % of the estimate.
It assumes the model is perfect, so it says nothing about whether the model or fit is good. Packages compute it differently, and it does not replace real repeatability studies.
24

Why is filtering results by %CRLB risky for low-concentration metabolites?
A low concentration gives a large %CRLB even when the absolute uncertainty is fine. Rejecting values above a %CRLB threshold removes low estimates preferentially → the group mean is biased upwards.
Better: compare with the typical CRLB in a normal cohort.
25

What do fit residuals and the FQN tell you?
Residual = data − fit; it should look like pure noise. Structure in it means something is missing from the model (a metabolite, the right lineshape or baseline, an artefact).
FQN (fit quality number) = residual variance / noise variance: ≈ 1 means the fit explains everything but noise.
26

Three ways to turn fitted areas into concentrations
1) External reference: phantom of known concentration (correct for coil loading and B1).
2) Internal ratio to tCr, NAA or Cho: simple, scale cancels, but the reference may itself change with disease.
3) Internal water reference: unsuppressed water from the same voxel, corrected for T1, T2, water content (and small MT effects). Units: mmol/kg wet weight.
27

Water-referenced concentration: the formula and why short TE + long TR help
[M] = (S_M/S_W) × (2/n_M) × [W] × (R_W/R_M), with R = (1 − e^(−TR/T1))·e^(−TE/T2) for each signal. 2 = protons in water, n_M = protons in the metabolite peak, [W] = 55.5 M × tissue water content.
Very short TE → T2 factors ≈ 1; fully relaxed long TR → T1 factors ≈ 1.
28

Worked: NAA area is 1/3000 of the water area. Fully relaxed, very short TE, water content 0.8. What is [NAA]?
[W] = 0.8 × 55,500 ≈ 44,400 mmol/kg. NAA's singlet has 3 protons, water 2.
[NAA] = (1/3000) × (2/3) × 44,400 ≈ 9.9 mmol/kg.
29

What is MRSI, and how does it encode position?
Any MRS sequence plus phase-encoding gradients: a spectrum in every voxel of a grid → metabolite maps (often ratios like [Tau]/[tCr]).
No readout gradient (the readout records the spectrum). Each phase step adds a position-dependent phase to each voxel's whole spectrum: FT over time → ppm, FT over the steps (k) → position.
Advantage: many regions compared in one scan; also works for 31P, 2H, functional and diffusion-weighted MRSI.
30

What makes MRSI harder than single-voxel MRS?
1) Low metabolite concentrations → low SNR, long scans.
2) Shimming a whole slice: B0, linewidth and frequency vary voxel to voxel.
3) Water suppression over the whole organ: one suppression frequency, but water's frequency varies (inhomogeneity, drift) → uneven suppression.
4) Lipids leak in through the point-spread function.
5) Hardware limits and in-house sequences.
31

MRSI with 3, 2 or 1 RF pulses: what does each select?
3 pulses (PRESS/STEAM/SPECIAL-MRSI): a box inside the brain, split into a grid → "square maps"; lipids avoided, but no full-brain coverage.
2 pulses (slice-selective spin echo): a slice.
1 pulse (FID-MRSI): a slice → whole-slice coverage, but strong scalp lipids need saturation bands.
32

Why is there no chemical-shift displacement along phase-encoded directions in MRSI?
The phase a gradient gives depends only on position, not on the metabolite's frequency, so all metabolites of a voxel land in the same place.
Trap: CSDE still exists along directions defined by slice-selective pulses (the slice, or the edges of a PRESS box).
33

How long does MRSI take compared with single-voxel MRS, and how is it accelerated?
T_MRSI = N_avg × Nx × Ny (× Nz) × TR; T_SVS = N_avg × TR (~5 min human, 10–15 min preclinical).
At TR 2 s: 32 × 32 = 34 min; 16 × 16 × 16 = 136 min.
Faster: short TR (low flip angle), several k-points per TR (spatial–spectral encoding), undersampling (parallel imaging, compressed sensing: AF 4, 119 → 28 min), prior-knowledge reconstruction.
34

Why do tiny MRSI voxels still get usable SNR?
Every excitation collects signal from the whole excited volume, and every phase-encode step contributes to every voxel.
So a 32 × 32 scan with one average acts like 1024 averages of each voxel. Compare methods with SNR per unit time = SNR/√(total time).
35

How do you reduce lipid leakage through the MRSI point-spread function, and why prefer acquisition weighting to a Hamming filter?
Saturation bands, more phase-encode steps, k-space weighting, lipid suppression in processing.
A Hamming filter applied afterwards cuts the PSF ripples but broadens the PSF (lower resolution) and down-weights high-k data you already paid for (less sensitivity per unit time). Acquisition weighting measures central k-space more often, shaping the PSF without that waste.
36

FID-MRSI: settings, pros and cons
One slice-selective pulse, phase encoding, FID after an ultra-short acquisition delay (~1–1.3 ms); short TR (200–800 ms), so a low flip angle (Ernst angle: most signal per unit time).
Pros: negligible J-evolution and T2 loss; fast (31 × 31, TR 811 ms ≈ 13 min at 14.1 T), so more phase steps → better PSF; whole-slice coverage.
Cons: the delay gives a first-order phase; whole-slice lipids need strong saturation.
Compared with PRESS-MRSI: more brain coverage even after QC; PRESS-MRSI gave slightly lower concentrations (possibly metabolite T1/T2 not corrected).
37

Saturation bands for FID-MRSI: why many thin bands hugging the skull?
FID-MRSI excites the whole slice, scalp lipids included, so bands must remove lipid all around the brain.
A few thick bands (e.g. six slabs, 4–12 mm) are simple but cut into the cortex. Seven thinner slabs (1.5–4 mm) placed tight against the skull, with some pairs overlapping on the same region for stronger saturation, remove lipid while keeping more cortex: more brain coverage.
38

MRSI with a surface coil: why does the slice orientation matter (axial vs coronal)?
A surface coil's sensitivity falls with distance from the loop.
A slice perpendicular to the coil spans many depths: only voxels near the coil give good signal, so the usable map is small (the slide's axial water map).
A slice parallel to the coil stays at one depth: sensitivity is even across it, so the map is larger and more uniform, with better coverage (the slide's coronal water map). Rule: put the MRSI slice parallel to the coil.
39

MRSI quality control: which metrics, and how are bad voxels handled?
Four metrics per voxel: SNR, linewidth (FWHM), B0 shift, CRLB.
Check the maps before fitting (no masking), then mask voxels that fail after fitting. MRS4Brain toolbox criteria: SNR above 75% of the mean, FWHM below 125% of the mean, CRLB < 40% (one slide example uses 30%).
Interpolating the displayed map only makes it look smoother; like zero-filling, it adds no resolution.
40

MRS4Brain toolbox: what is it and what does its MRSI pipeline do?
A free EPFL MATLAB toolbox for MRSI, single-voxel and diffusion-weighted spectroscopy.
MRSI: load Bruker data → brain mask (water signal) → HSVD water removal → lipid suppression → quick maps (linewidth, ΔB0, SNR = NAA height/σ noise) → LCModel fit of every voxel → QC thresholds (SNR, FWHM, CRLB) → register to an anatomical template + atlas (SIGMA) → metabolite maps and region tables.
41

How does MRS4Brain suppress lipids in FID-MRSI?
It builds a set of typical lipid spectra from the scalp and removes from each brain voxel whatever looks like them (SVD-based, with L2 regularisation: a penalty weight balances removing lipid against keeping metabolite signal).
It works alongside saturation bands and finer grids, not instead of them.
42

"Fill gaps" (in development in MRS4Brain): what problem does it solve?
FID-MRSI starts recording after an acquisition delay, so the first FID points are missing. That missing start is what gives the spectra a first-order phase.
Fill gaps rebuilds those points, removing the problem at its source instead of correcting the phase afterwards.
43

LCModel baseline knot spacing (DKNTMN): what does it control?
The baseline is a smooth spline with knots this far apart (demo: 0.25 ppm). Larger spacing = stiffer, smoother baseline.
Too flexible: the baseline swallows part of the broad metabolite multiplets, which are then underestimated. Too stiff: it cannot follow real broad humps (macromolecules, lipids).
44

LCModel phase settings (DEGZER ± SDDEGZ, DEGPPM ± SDDEGP)
The expected zero-order phase (degrees) and first-order phase (degrees/ppm), each with how far the fit may move from it.
Demo: zero-order 0 ± 999° (free), first-order 0 ± 0 (held fixed).
45

LCModel WCONC = 44 444 mM and "relative metabolite Cr+PCr = 8 mM": what are they?
WCONC is [W] in the water-referencing equation: 55.5 M × 0.8 water content ≈ 44 400 mmol/kg (the slide's "44.444 mM" is a slip).
"Relative" maps instead scale every metabolite to an assumed tCr of 8 mM: a ratio with a familiar unit, so it inherits the ratio caveat (tCr itself might change).
46

Region-wise MRSI results: how are voxels assigned to brain regions?
The anatomical MRI is registered to a template with an atlas segmentation, so each MRSI voxel gets region labels (left, right or both).
An overlap threshold sets how much of an MRSI voxel must lie in the region: low = more voxels but more contamination from neighbours (partial volume); high = purer but fewer voxels (demo: 0.8).
47

MRSI concentration table: what does the "N voxels" column tell you?
How many voxels in the region passed QC for that metabolite. It shows how reliable the mean is.
Demo: NAA, Ins, Tau, Glu in all 144 voxels, but Ala 17, Lac 21, GPC 72. Those means come from the few voxels where the fit was confident, which are biased towards high values (the %CRLB trap).
Réviser les 47 cartes gratuitement

Plus de flashcards : Médecine et santé

Biology and Cognition: Visual System75 cartesMédecine et santé · AnglaisMpharm tech exam: cardiac brand & generic names59 cartesMédecine et santé · AnglaisZoyaBlood49 cartesMédecine et santé · AnglaisBobertRutgers - Biology 2 - Endocrine 239 cartesMédecine et santé · AnglaisMariaBlood groups and transfusions31 cartesMédecine et santé · AnglaisfaithInflammation25 cartesMédecine et santé · AnglaiscgilnursingMastication and Face20 cartesMédecine et santé · Anglaisjocyt9funciones rrhh parcial15 cartesMédecine et santé · EspagnolAbrilProduits sanguins labiles (PSL)11 cartesMédecine et santé · FrançaisClaraCardiac Arrest Pathophysiology and Resuscitation168 cartesMédecine et santé · AnglaisninaPharmacology: Lisinopril and Losartan144 cartesMédecine et santé · AnglaisVictorAnatomy and Physiology: Musculoskeletal System100 cartesMédecine et santé · AnglaisExalted

Tous les decks : Médecine et santé

Révise ces decks dans l’app

Gratuit sur iPhone, Android, Mac et le web. Ajoute n’importe quel deck à ton compte en un tap.

Télécharger dans l’App StoreDisponible sur Google Play
Réviser les 47 cartes gratuitement