14.2 MRI and CT/MR Perfusion

MRI and CT/MR Perfusion

Quick Answer: DWI bright plus ADC dark = cytotoxic restriction (hyperacute infarct). FLAIR and T2 show edema later. GRE/SWI bloom blood. CTP/MRP split core (collapsed CBF/CBV) from penumbra (Tmax delay). DEFUSE-3 style mismatch is core ≤70 mL and ratio ≥1.8—an imaging concept, not this guide’s stroke-treatment chapter. MRI beats CT in the posterior fossa. Skip the magnet if the device is unsafe, the patient is crashing, or a faster CT stroke protocol is the clock you have.

Independent OpenExamPrep teaching in this section covers MRI and CT/MR perfusion listed under Diagnostic studies and procedural skills in the ABPN Content Specifications (02.A.2–3). This guide is not an ABIM or ABPN product.

Sequence toolkit in the neuro ICU

You do not order a generic MRI. You order a question. Each contrast mechanism answers a different one.

Bold the ADC. A stroke stem that says DWI bright without ADC is incomplete. True restriction is DWI hyperintense and ADC hypointense. T2 shine-through (chronic infarct, some vasogenic edema) is bright on DWI and bright or normal on ADC. Treat shine-through as old water, not as a new core.

DWI/ADC in hyperacute ischemia

Cytotoxic edema slows water diffusion within minutes. DWI can be positive by about 30 minutes from onset, far earlier than NCCT hypodensity or T2 change. That is why the previous section said a normal 90-minute CT does not exclude infarct.

Worked example: last seen well 70 minutes ago, NIHSS 14, NCCT ASPECTS 10. DWI shows a bright left insula and ADC is dark in the same pixels. That is hyperacute MCA ischemia, not a chronic leukoaraiosis patch (chronic patches are T2/FLAIR bright with high ADC).

FLAIR often remains nearly normal in the first ~4.5 hours. A DWI-positive, FLAIR-negative pattern is the imaging idea behind unknown-onset / wake-up protocols (WAKE-UP used DWI–FLAIR mismatch). FLAIR-positive cortex in the same territory means the infarct is old enough to have accumulated T2 water—useful chronology, still not a full treatment algorithm.

DWI is not exclusive to arterial infarct. Abscess, Creutzfeldt–Jakob, seizure-related cortical change, hypoglycemia, and highly cellular tumor can restrict. Clinical context and ADC extent keep you from calling every bright DWI a candidate for lytics.

T2 edema versus GRE/SWI blood

T2 (and FLAIR) show edema: vasogenic (tumor, abscess, PRES—bright T2, facilitated ADC) versus cytotoxic (infarct—bright T2 later, restricted ADC early). Mass effect from T2-bright swelling in a malignant MCA infarct is how you anticipate herniation, but the earliest infarct signature remains DWI, not T2.

Gradient recalled echo (GRE) and susceptibility-weighted imaging (SWI) are the MRI blood sequences. Paramagnetic blood products distort the local field and bloom. SWI is more sensitive than GRE for microbleeds, diffuse axonal injury shear hemorrhages at the gray–white junction and corpus callosum, and the susceptibility vessel sign of thrombus. A CT-negative, MRI-positive convexity SAH or a field of amyloid microbleeds is an SWI diagnosis.

Blood on conventional T1/T2 follows hemoglobin state (oxy → deoxy → intra- then extracellular methemoglobin → hemosiderin). For ICU decision-making, remember the practical ends: GRE/SWI dark bloom for acute and chronic blood products, T1 bright for subacute methemoglobin, T2 dark rim for chronic hemosiderin. Do not use a non-contrast T1-only survey as your hemorrhage screen.

Posterior fossa: MRI’s home field

CT beam-hardening through the petrous bones hides the brainstem, middle cerebellar peduncles, and fourth ventricle. A lateral medullary or small pontine infarct that will declare itself as an NIHSS of 3 and an aspiration risk can be CT-negative and DWI-positive. Cerebellar infarct that will swell at 48–72 hours is likewise more honest on MRI. If the stem is vertigo, crossed findings, or a “normal CT, still can’t walk,” the imaging answer is MRI with DWI, not a third NCCT.

Contraindications: device, instability, time

MRI is the wrong first test when any of the following dominate:

Gadolinium is unnecessary for DWI diagnosis of infarct. If contrast MRI is required in advanced CKD, prefer group II agents; NSF is the group I / very low GFR story. Pregnancy is a “need-to-know” conversation, not a default contrast MRI.

CT and MR perfusion: core versus penumbra

CT perfusion (CTP) tracks an iodinated bolus with repeated cine CT and builds time–density curves. MR perfusion (MRP) most often uses dynamic susceptibility contrast after gadolinium; arterial spin labeling (ASL) can estimate CBF without contrast. Both try to separate tissue that is already dead from tissue that is ischemic but still salvageable.

Operational definitions used on automated CTP maps (RAPID and similar) commonly call core the volume with relative CBF <30% of contralateral brain, and hypoperfusion / penumbra the volume with Tmax >6 seconds. Mismatch volume is hypoperfusion minus core. Mismatch ratio is hypoperfusion volume divided by core volume.

Pitfalls: very early CTP can overestimate core (ghost core). Seizure, chronic stenosis, and poor cardiac output scramble maps. Tiny cores with huge delays still need a vessel occlusion on CTA/MRA to make physiologic sense. Perfusion is a selection tool, not a stand-alone diagnosis of stroke.

DEFUSE-3 style mismatch, conceptually

DEFUSE-3 (Albers et al., NEJM 2018) tested endovascular therapy 6–16 hours after last known well in anterior-circulation large-vessel occlusion using perfusion imaging. The imaging profile that defined a favorable mismatch was:

• Ischemic core ≤70 mL
• Mismatch ratio ≥1.8 (Tmax >6 s volume / core)
• Mismatch volume ≥15 mL

Worked example: core 22 mL, Tmax >6 s volume 95 mL. Mismatch volume = 95 − 22 = 73 mL. Ratio = 95 / 22 ≈ 4.3. That is a favorable imaging mismatch on DEFUSE-3 rules. A core of 95 mL with Tmax volume 110 mL (ratio 1.16) is not.

DAWN used a different idea: clinical–core mismatch (NIHSS versus core volume, with age bands) out to 24 hours. You will meet both trials in the acute ischemic stroke chapter. This section’s job is to read the maps: small core, large Tmax lesion, ratio at least 1.8. Do not turn this chapter into a thrombectomy consent lecture, and do not quote DEFUSE-3 thresholds for a cerebellar perforator with no vessel imaging.

Malignant profile teaching: a huge core with little penumbra means little to salvage and more reperfusion hemorrhage risk. Large-core thrombectomy trials exist and are a treatment-chapter nuance; they do not erase the need to measure core and penumbra honestly.

Putting MRI and perfusion next to CT

If GFR is low, DWI and TOF MRA still work without iodine or gadolinium. CTP is an iodine bolus; MRP is usually a gadolinium bolus. Choose the contrast that the kidneys and the clock can tolerate.

Exam-style traps

• Calling DWI-bright, ADC-bright tissue an acute core (that is shine-through)
• Using T2 without DWI as the hyperacute stroke sequence
• Sending an MRI-unsafe pacemaker or an actively herniating patient to the magnet as first imaging
• Equating any Tmax delay with core
• Treating DEFUSE-3 numbers as if this were the mechanical thrombectomy chapter rather than an imaging-selection concept
• Trusting CT over MRI for a crossed brainstem syndrome

Independent practice items at /practice/abim-neurocritical-care are a study bank for this reasoning, not the computer-based examination administered by ABPN.