8  MR Spec

8.1 Single vs Multivoxel

In MR spectroscopy, what is single voxel vs multi voxel technique. Please explain.

8.1.1 Overview of MR Spectroscopy (MRS)

Magnetic-resonance spectroscopy extends routine MRI by measuring signal frequency rather than magnitude. Frequency shifts created by the chemical environment of nuclei (typically ^1H, occasionally ^31P, ^13C, ^19F, etc.) let us quantify tissue metabolites in vivo. Two acquisition strategies dominate clinical brain work:

Acquisition strategy Synonyms
Single-voxel spectroscopy (SVS) “Single-voxel MRS”, PRESS/STEAM
Multi-voxel spectroscopy “Chemical-shift imaging (CSI)”, “Spectroscopic imaging”, 2-D or 3-D MRSI

8.1.2 Single-voxel Spectroscopy (SVS)

  • How it works – Three orthogonal slice-selective RF pulses (PRESS) or one 90° + two 180° pulses (STEAM) excite a single cuboid volume (typically 1–8 cm³).

  • Data – One composite spectrum representing the average metabolic profile of that voxel.

  • Strengths

    • Highest possible signal-to-noise ratio (SNR) per unit time.
    • Short scan time (1–3 min).
    • Straight-forward shimming and post-processing; robust even on mid-range scanners.

  • Limitations

    • No spatial heterogeneity information.
    • Placement errors or partial-volume with CSF/necrosis can mislead.
    • Must repeat the sequence for each additional location.

Typical use-cases: Focused evaluation of a clearly visible mass, seizure focus, or deep-seated nucleus; longitudinal follow-up of a treated lesion using identical voxel coordinates.

8.1.3 Multi-voxel Spectroscopy / Chemical-Shift Imaging (CSI)

  • How it works – An excitation slab is followed by phase-encoding gradients in 1, 2, or 3 orthogonal directions, producing a grid of spectra (e.g., 16×16×1 or 16×16×8) within the FOV. Each voxel may be 0.5–1 cm³.

  • Data – Hundreds–thousands of spectra enable metabolite maps (heat-maps of N-AA, Cho/Cr ratio, etc.) co-registered with anatomic MRI.

  • Strengths

    • Depicts spatial distribution and heterogeneity—helpful for tumor grading, treatment-response mapping, epilepsy surgical planning, etc.
    • More efficient when many regions must be examined simultaneously (e.g., bilateral hippocampi).

  • Limitations

    • Lower SNR per voxel – often requires longer acquisition (5–15 min) and/or larger voxels.
    • Demands excellent B₀ homogeneity over a wide FOV; prone to lipid bleed-over, susceptibility, fold-over artifacts.
    • Heavier reconstruction and interpretation workload.

Typical use-cases: Pretreatment glioma evaluation (identify most aggressive region for biopsy), differentiating tumor recurrence vs. radiation necrosis, mapping metabolic penumbra in stroke or traumatic brain injury.

8.1.4 Summary Table

Feature Single-voxel (SVS) Multi-voxel / CSI (MRSI)
Spatial coverage One user-defined cuboid (1–8 cm³) 2-D or 3-D grid covering large slab/volume
SNR per voxel High Lower (shared time & flip-angle across voxels)
Acquisition time Short (1–3 min) Longer (5–15 min)
Information obtained Average spectrum of target voxel Spectrum for every voxel → metabolite maps
Shimming & processing Simple, robust More demanding; post-processing pipeline needed
Artifacts Minimal; mostly motion & partial-volume B₀ inhomogeneity, lipid contamination, chemical-shift displacement
Best for Focal lesions, follow-up at identical coordinates, deep nuclei Tumor heterogeneity, biopsy planning, metabolic mapping, multifocal disease
Scanner/software requirements Available on virtually all systems Requires higher-order shims, advanced spectroscopy package

8.1.5 Practical Tips for Selecting a Technique

  1. Clinical question & lesion extent

    • If you need metabolic data from one well-defined focus → SVS.
    • If you must explore where metabolism is abnormal or pick the hottest biopsy target → CSI.

  2. Time & patient factors

    • Uncooperative or critically ill patient → shorter SVS often succeeds where CSI fails.

  3. Hardware quality

    • On older 1.5 T systems with limited shims, SVS reliability > CSI.
    • High-field (3 T+) and modern gradient systems improve CSI quality substantially.

  4. Combine when necessary

    • A hybrid workflow is common: start with CSI for whole-lesion overview; add high-SNR SVS over the most suspicious voxel for therapy monitoring or research quantification.
Bottom line:

Single-voxel MRS gives you a fast, high-quality snapshot of metabolism in one place, whereas multi-voxel CSI paints a metabolic map across the brain at the cost of longer scans, more complex setup, and lower SNR per voxel.

8.2 Short TE vs Long TE

In my institution, they use Single-voxel MRS in TE = 144 and TE = 35 for the brain tumor. Why use different TE, what are the rational and physics?

8.2.1 Why a brain-tumor protocol often acquires two single-voxel spectra

(Short-TE ≈ 35 ms and Intermediate-TE ≈ 144 ms)

Parameter Short TE ≈ 35 ms Intermediate / “long” TE ≈ 144 ms
T2 filtering Minimal – preserves signals from short-T2 metabolites (mIns, Glx, mobile lipids, macromolecules) Stronger – broad macromolecular background and very-short-T2 lipids are attenuated, giving a flatter baseline
J-coupling evolution Early in the J-evolution curve; coupled spins are in-phase → lactate doublet sits upright and can be confused with lipid TE ≈ 1 / J for lactate (J≈7 Hz → 1/(2 J)≈70 ms; odd multiple → ~140 ms) → coupled spins acquire 180° phase → the lactate doublet is inverted below the baseline, confirming its identity
Metabolites best seen mIns (glial marker), Glu/Gln, GSH, mobile lipids (necrosis), combined Lac/Lip, scyllo-inositol, glycine Cho, Cr, NAA (cleaner quantification), pure lactate, alanine (meningioma), amino acids (abscess)
Signal-to-noise (per unit time) Highest—little T2 decay Lower—T2 decay and filtration reduce amplitude
Clinical questions helped Detect lipid peaks that suggest necrosis/high-grade tumor or abscess; pick up mIns elevation that favors glioma over lymphoma (PubMed) Confirm or rule out lactate → anaerobic metabolism, tumor aggressiveness or treatment effect; easier Cho/NAA or Cho/Cr ratios
Typical pitfalls Huge baseline makes automatic fitting harder; Lac vs Lip overlap Loss of mIns & Glx can hide low-grade glioma clues; SNR penalty in small voxels

8.2.1.1 Underlying Physics

  1. Echo time (TE) and T2-decay After localization (PRESS or STEAM), transverse magnetisation decays as e^{-TE/T2}.

    • Short TE (≈ 35 ms) ≪ T2 of most metabolites (100–300 ms) → minimal decay, so even broad mobile lipids/macromolecules with T2 ≈ 30–70 ms survive.
    • At TE ≈ 144 ms, those short-T2 components are heavily damped, flattening the baseline and making narrow resonances (Cho, Cr, NAA) stand out.

  2. J-coupling (spin–spin scalar coupling) Coupled spins oscillate between in-phase (doublet upright) and antiphase (doublet inverted) every 1/J seconds.

    • Lactate has J≈7 Hz → period ≈ 143 ms. Thus at TE = 144 ms the CH₃ doublet at 1.33 ppm is inverted; at TE = 35 ms it is upright and can imitate lipid. This phase behaviour is a time-honoured trick to prove that an apparent 1.3-ppm peak is truly lactate (PubMed Central, Barrow Neurological Institute).

  3. Macromolecular baseline interference Short-TE spectra contain a broad background from proteins and cytosolic lipids that can bias fitting; increasing TE suppresses these unmodelled components and improves reproducibility of Cho/NAA ratios (PubMed Central).

8.2.1.2 Practical Rationale in a Tumor Study

  • Start with TE ≈ 35 ms

    • Collect the most complete metabolic profile (lipid burst, mIns, Glx) with maximal SNR.
    • Lipid ± lactate complex at 0.9–1.3 ppm helps separate necrotic high-grade glioma or abscess from low-grade lesions.

  • Add TE ≈ 144 ms

    • Clean spectrum for routine ratios (Cho/Cr, Cho/NAA) that correlate with cellular proliferation.
    • Confident detection of an inverted lactate doublet indicates anaerobic glycolysis (high-grade tumour, treatment-related necrosis) (PubMed).
    • Disappearance of broad lipid at long TE but persistence at short TE differentiates mobile lipids (short T2) from lactate (longer T2).

By combining the two echo times, your protocol balances sensitivity (short TE) with specificity and quantification clarity (long TE), maximising the diagnostic information from a single-voxel examination while adding only a minute or two of scan time.

Key takeaway: TE controls both T2-weighting and the phase evolution of J-coupled spins. A short TE (~35 ms) spectrum is a wide-net capture of all metabolites—including those with short T2—whereas an intermediate TE (~144 ms) acts as a biochemical “filter” that flattens the baseline and flips lactate, giving you a cleaner, more interpretable signature of tumour metabolism.