How to read a head MRI is a systematic look at the sequences, the planes, and a short list of things that actually change management: stroke, mass effect, demyelination, blood products, and the artifacts that mimic them. It is not MRI physics as a Tesla explainer. It is not “what a cross-section is.” It is not a PYCAD MRI-reader product, and it is not a substitute for a radiologist.
If you meant spatial / temporal / contrast resolution, 1.5 T vs 3 T → resolution of MRI (a later leftover; not rewritten here). If you meant what a medical cross-section is → cross-section analysis. If you meant liver segments → those URLs are 688, not this page. If you meant the viewer you open the study in → DICOM viewer software.
PYCAD builds custom web DICOM viewers and imaging models. That is how a head MRI opens in a clinic app. It is not a “read this scan” service.
The sequences are the language
Each series is a different physics experiment on the same head. You do not diagnose from one. You compare how a region behaves across them. The words are simple: hyperintense (bright) and hypointense (dark). The useful fact is the pattern across the set.
| Sequence | What it is for | CSF | Fat | Typical use on a head study |
|---|---|---|---|---|
| T1-weighted | Anatomy. Gray / white matter | Dark | Bright | Structure. Post-contrast enhancement (gadolinium) |
| T2-weighted | Water: edema, inflammation, many tumors | Bright | Bright | The first “something is wet” look |
| FLAIR | T2 with CSF suppressed | Dark | Bright | Periventricular and cortical lesions that T2 would hide in bright CSF |
| DWI (+ ADC) | Restricted water motion | Variable | Dark | Acute infarct. Abscess. Some dense tumors. Always read with the ADC map |
| GRE / SWI | Susceptibility: blood, calcium, metal | Variable | Variable | Hemorrhage, microbleeds, some vascular malformations |
An acute infarct is the teaching example: bright on DWI because water is trapped in swollen cells, dark on the matching ADC. In the first hours it can be invisible on T2. One sequence is a clue. The set is the read.
How sharp the pixels are — field strength, matrix, slice thickness — is 572, not this page. A blurry study is a quality problem, not a different search pattern.
A search pattern, every time
The failure mode is “search and satisfy”: you see the obvious lesion and stop. Build a path and walk it on every study, including the ones that look normal.
- Right patient, right date, right side. Then: is the study complete, and is motion ruining it?
- A global pass on axial T2 and FLAIR. Hemispheres, sulci, ventricles — is the picture lopsided before you zoom?
- Outside in. Scalp, skull, extra-axial spaces (blood, pus, a mass that is not brain). Then ventricles and sulci (hydrocephalus vs effacement vs atrophy). Then cortex and white matter, lobe by lobe, left vs right. Then deep gray (basal ganglia, thalami). Then the posterior fossa: midbrain, pons, medulla, cerebellum. People skip the last one.
Link the series in the viewer and scroll them together. Window as you go. That is a hanging-protocol job, not a slogan. Planes (axial / sagittal / coronal) as a concept are 689.
Four patterns worth a name
Acute ischemic stroke
DWI bright, ADC dark, in a vascular territory. That mismatch is the acute fingerprint. An old infarct is a hole: dark on T1, bright gliosis on T2/FLAIR, no DWI/ADC mismatch. “Time is brain” is why DWI is early in the stack when the history is sudden.
Mass effect
A tumor (or a bleed, or edema) is often found by what it does to neighbors: a squashed ventricle, a midline shift, sulci that have been ironed flat against the skull. T2/FLAIR show the edema. Post-contrast T1 shows where the blood–brain barrier is broken — if contrast was given. Enhancement is a clue, not a diagnosis.
Demyelination (MS as the type case)
Periventricular ovoid T2/FLAIR lesions — “Dawson’s fingers” when they point away from the ventricle — are the textbook look. FLAIR exists so those plaques are not lost in bright CSF on T2. This page is the pattern, not an MS clinic.
Blood products after trauma
A large extra-axial collection is hard to miss. Diffuse axonal injury is not: GRE/SWI microbleeds at the gray–white junction, or in the corpus callosum and brainstem. Those dark dots are sheared vessels. They change the story of a “normal” CT.
Artifacts and things you were not asked about
- Motion. Ghosts and smears that ignore anatomy. Repeat if you can; do not call a streak a lesion.
- Susceptibility. Dark voids at air–bone (sinuses) and around metal. Worst on GRE/SWI. Ask whether the “bleed” sits exactly where dental work or a shunt lives.
Incidental findings are real anatomy you were not sent to find: a small arachnoid cyst that follows CSF on every sequence, a few nonspecific white-matter dots in an older adult, a simple pineal cyst. Published series disagree on how often they appear; the clinical job is the same — is it mass effect, is it growing, does it answer the question on the order? A range invented for a blog is not a prevalence study. One open-access review of incidental brain findings is PMC9975529.
The report
Impression first, in the language of the order (“no acute infarct,” not a tour of every sequence). Findings in the same outside-in order you used. Technique in one line. Hedge when the evidence is thin; do not hedge a DWI/ADC mismatch you are sure of. This is communication, not a PYCAD reporting product.
What this page is not
- Not 572. Tesla, matrix, and “what resolution means” wait their leftover pass.
- Not 689. The slice / plane survey already exists.
- Not 688. Liver is not a head.
- Not a second reader, not an AI diagnosis page, not a PYCAD “MRI reader.” Dropped Outrank stills, a YouTube embed, a portfolio CTA, and an unsourced 86.3% accuracy claim.
If the missing piece is a web viewer that can hang T1 / T2 / FLAIR / DWI / ADC together, that is the imaging piece. Case studies.