MR Perfusion & Spectroscopy
- Perfusion MRI measures blood flow through tissue; spectroscopy measures the chemistry of tissue. Different questions, different tools.
- The headline perfusion number is relative cerebral blood volume (rCBV) — high rCBV means a lot of vessels, which screams aggressive tumor (or tumor recurrence, not radiation scarring).
- MR spectroscopy (MRS) reads out metabolite peaks. The cast of characters: NAA (healthy neurons), choline (cell turnover), creatine (the reference), lactate/lipid (things going badly).
- The classic tumor signature: choline up, NAA down — sometimes summarized as a "Cho/NAA flip."
- Both are problem-solvers, not screening tools. You order them to answer a specific question the standard sequences couldn't.
So you've done your standard brain MRI — the diffusion, the T2, the post-gadolinium sequences — and you still can't tell whether that blob is an aggressive tumor, a sleepy one, an abscess, or last year's tumor coming back to ruin everyone's day. This is where perfusion and spectroscopy earn their keep. They don't just photograph the lesion; they interrogate it.
Perfusion: how busy are the pipes?
Standard MRI shows you the plumbing's shape. Perfusion shows you the flow — how much blood is actually moving through a chunk of tissue, and how fast.
The most common flavor is DSC (dynamic susceptibility contrast). You inject a bolus of gadolinium and then scan very fast as it races through the brain. As the contrast packs into the capillaries, it briefly distorts the local magnetic field and the signal dips — like a crowd of people momentarily darkening a doorway as they rush through. The deeper and tidier that dip, the more blood volume in that voxel. From that signal-versus-time curve we estimate relative cerebral blood volume (rCBV), the workhorse number.
There's more than one way to skin this cat. DSC watches signal drop as contrast passes through. DCE (dynamic contrast-enhanced) watches signal rise over a longer window and leans toward leakiness/permeability. And ASL (arterial spin labeling) uses magnetically tagged blood as its own tracer — no gadolinium at all. Same goal, different bookkeeping.
Why do we care? Because aggressive tumors build their own messy, leaky vasculature to feed themselves — lots of vessels means high rCBV. Low-grade lesions tend to stay quieter.
High rCBV = lots of blood volume = think aggressive, vascular, high-grade. It's the single number most people reach for first.
The killer application is the question every neuroradiologist dreads: tumor recurrence vs. radiation injury. After treatment, both can light up with contrast on a standard scan and look maddeningly identical. But recurrent tumor regrows its hungry vasculature (high rCBV), while radiation necrosis is dead, poorly perfused tissue (low rCBV). Perfusion can break the tie when the plain pictures shrug.
Spectroscopy: reading the tissue's chemistry
If perfusion asks "how much blood?", spectroscopy asks "what is this tissue made of?" MR spectroscopy (MRS) ignores anatomy and instead plots a little graph of chemical peaks from a chosen voxel — basically a chemistry readout of one small box of brain.
Think of it as the tissue's pantry inventory. The main shelves:
| Metabolite | What it signals | What "up" or "down" usually means |
|---|---|---|
| NAA (N-acetylaspartate) | Healthy, intact neurons | Down = neurons lost or pushed aside (tumor, injury) |
| Choline (Cho) | Cell membrane turnover | Up = lots of cells being built (tumor) |
| Creatine (Cr) | Energy metabolism | Fairly stable — used as the reference to compare against |
| Lactate | Anaerobic metabolism | Up = the tissue has run out of oxygen |
| Lipid | Breakdown of cell membranes | Up = necrosis, dead tissue |
The pattern that should make a tumor jump to mind: choline climbs, NAA falls. Healthy neurons (NAA) get crowded out while membranes turn over like crazy (choline) to build new tumor cells. People shorthand this as the "Cho/NAA flip" — the two peaks trade places compared to normal brain.
Putting them to work (and where they bite back)
These tools shine on the hard differentials: grading a glioma, distinguishing tumor from a mimic, or sorting recurrence from treatment effect after radiation. A few patterns are genuinely useful — for example, a ring-enhancing lesion that's a brain abscess often shows characteristic markers like amino acids on MRS, which a necrotic tumor won't.
Perfusion and spectroscopy are noisy. A spectroscopy voxel parked over hemorrhage, calcium, fat, or near bone gives garbage peaks. Perfusion rCBV gets faked-out by big vessels, hemorrhage, and contrast leaking through a broken blood-brain barrier. Always read these alongside the conventional sequences — never let one funny-looking number overrule the whole picture.
These are problem-solvers, not screeners. You don't sprinkle them on every brain MRI. You add them when a specific question — grade? recurrence? abscess vs. tumor? — can't be answered from the standard images alone.
The mindset shift is the whole point: conventional MRI tells you where and how big. Perfusion tells you how vascular, and spectroscopy tells you what it's made of. Stack those three answers together and a lesion that was a mystery on the plain scan often confesses.