Subdural vs Extradural Haematoma: What Actually Happens in the ER
Most people think these two are easy to tell apart on CT. They're not. I spent years reading head trauma scans and can still get tripped up when the bleed is atypical. The classic textbook descriptions are fine for exams but the real world doesn't always cooperate. An extradural haematoma (EDH) sits between the skull and the dura mater. It's almost always arterial – usually the middle meningeal artery tearing from a temporal bone fracture. The blood pushes the dura away from the skull, creating that classic lens shape on CT. Biconvex is the word you want. Fast, bright, and usually dramatic in presentation. The patient hits their head, has a lucid interval, then deteriorates rapidly. A subdural haematoma (SDH) is between the dura and the arachnoid. These are venous – bridging veins shearing off as the brain moves inside the skull. That's why they look like a crescent, following the curve of the brain rather than crossing suture lines. Slower onset usually, though acute SDH can present just as violently. In the elderly, even a minor bump from a few days ago can show up as a substantial collection by the time they actually come in.
The tricky part is when both features overlap or when the bleed isn't clean. I've seen SDHs that look lenticular because the blood is under enough pressure to push hard against the brain surface. And EDHs can appear crescentic when there's concurrent subdural bleeding or when the haematoma tracks along a fracture line in an unusual pattern. Here's something textbooks don't stress enough: the location matters more than the shape alone. Temporal region with a fracture line crossing the middle meningeal groove? Think EDH first. Across the convexity over the frontal lobe without a major fracture? SDH is more likely. But don't rule either out based on location alone. I remember one case that still bugs me. Patient came in after a motor vehicle collision. CT showed what looked like a massive SDH – broad crescent along the right hemisphere. Standard teaching said surgical evacuation. But on closer inspection, there was a thin fracture line through the temporal bone right where the middle meningeal artery runs, and the collection was actually tracking epidurally with a small subdural component. Pure SDH protocol would have missed the arterial source. We called neurosurgery earlier than we normally would have for a "crescent," and the surgeon confirmed it was predominantly an EDH. Turned out the venous component would have resolved; the arterial bleed needed clipping. Different patient outcome because we treated the right pathology. That one made me recheck every borderline case twice.
Another detail that catches people out: hyperacute SDHs can be isodense to brain on non-contrast CT, basically invisible until they cause mass effect. If you're looking at a scan from a patient who presented within hours of trauma and the CT seems "normal" but the clinical picture is wrong, go back and look harder. Subtle sulcal effacement, loss of gray-white differentiation on one side. Repeat imaging in six hours if you're unsure. EDHs rarely hide like this – they're usually bright from the start because arterial blood clots quickly and appears hyperdense almost immediately. The other thing nobody mentions enough is the dural attachment. EDHs can cross suture lines but they stop at dural reflections like the falx and tentorium. SDHs can cross sutures freely but they're limited by dural layers too. So a collection that extends across the midline in the posterior fossa is almost certainly subdural – the tentorium prevents an epidural from going there. A collection that stops dead at a cranial suture? More likely subdural as well, since EDHs tend to be constrained by the dura's attachment points rather than the bones themselves. Chronic SDHs deserve a separate conversation entirely. They're hypodense on CT, often with internal membranes and fluid-fluid levels from rebleeding. Management is completely different – burr hole drainage rather than craniotomy, and recurrence rates are significant. Don't mix up your treatment algorithms between acute and chronic presentations.
