How to Measure Intracranial Pressure (ICP)

Intracranial pressure is most accurately measured by placing a sensor or catheter directly inside the skull, with the external ventricular drain being the clinical gold standard. But that invasive approach is not always practical or safe, so clinicians also rely on a growing toolkit of less invasive and fully non-invasive methods, from bedside ultrasound of the eye to transcranial Doppler and even machine-learning algorithms that predict pressure spikes before they happen. The method chosen depends on how urgently the information is needed, how long monitoring must continue, and the risks a particular patient can tolerate.

Why Intracranial Pressure Needs Monitoring

Your skull is essentially a closed, rigid box. Inside it sit three things: brain tissue, blood, and cerebrospinal fluid (CSF). Because the skull cannot expand, those three components exist in a delicate balance: if one takes up more space, one of the others has to give way to keep pressure stable.¹ A brain bleed adds volume. A tumor adds volume. Swelling after a traumatic injury adds volume. When the compensatory mechanisms run out of slack, pressure climbs, and elevated intracranial pressure can compress brain tissue, cut off blood flow, and cause permanent damage or death.

Current management guidelines typically set the alarm threshold at around 20 to 22 mmHg, above which treatment should begin, and aim for a cerebral perfusion pressure in the range of 60 to 70 mmHg.² Knowing the exact pressure, rather than guessing from symptoms, lets clinicians intervene at the right time and avoid both under-treatment and unnecessary escalation.

External Ventricular Drains, the Gold Standard

An external ventricular drain (EVD) is a thin catheter threaded through a small hole in the skull into one of the brain’s fluid-filled ventricles. It serves a dual purpose: it reads pressure continuously and it can drain excess CSF or blood to relieve that pressure.³ The procedure is usually performed at the bedside by a neurosurgeon or neurointensivist, often under urgent conditions, using surface landmarks on the skull to guide placement.

The catch is that you cannot drain fluid and measure pressure at the same time. The drain must be temporarily clamped so the fluid column equilibrates and gives a true reading. In practice, many of those closure episodes are too short. One study of over 32,000 EVD closure episodes found that roughly two-thirds lasted less than one minute, and only about 23% met the criteria for full pressure equilibration before the drain was reopened.⁴ That means the pressure number being charted is sometimes a snapshot taken before the system had time to settle, which has obvious implications for treatment decisions based on those readings.

Parenchymal Sensors

When you need pressure data but do not need to drain fluid, a small sensor can be placed directly into the brain tissue (the parenchyma) instead of the ventricle. These devices are smaller and somewhat easier to position, though they still require a hole in the skull. Several sensor technologies are in use. Piezoelectric (strain gauge) sensors generate an electrical signal when pressure pushes on them. Fiber-optic sensors use a tiny mirror that shifts with pressure, changing the pattern of reflected light. Pneumatic sensors rely on a small air-filled balloon whose volume changes with surrounding pressure.⁵

Among these, piezoresistive strain gauge sensors tend to deliver the highest accuracy and the smallest drift over time. Fiber-optic sensors show promise for future applications but are more prone to calibration shift after several days inside the skull.⁶ Unlike EVDs, parenchymal sensors cannot drain fluid, so they trade therapeutic utility for simpler placement and potentially fewer mechanical complications.

Lumbar Puncture

A lumbar puncture, the familiar “spinal tap,” measures pressure in the fluid surrounding the spinal cord rather than inside the skull itself. For patients whose CSF pathways communicate freely (meaning there is no blockage between the brain’s ventricles and the lumbar space), the pressure measured at the lower back closely matches the pressure inside the brain. In patients with communicating hydrocephalus, one study found the average difference between lumbar and brain-tissue pressure was only about 0.75 mmHg, with near-perfect correlation between the two.⁷

This makes lumbar puncture a reliable diagnostic tool in conditions like idiopathic intracranial hypertension, where the question is whether pressure is abnormally high rather than whether it is changing minute to minute. It is not, however, suitable for continuous monitoring in the intensive care unit, and it can be dangerous if there is a mass or obstruction that could cause brain herniation when spinal fluid is withdrawn.

Optic Nerve Sheath Diameter Ultrasound

The optic nerve is wrapped in a sheath that is continuous with the membranes surrounding the brain. When intracranial pressure rises, that sheath balloons outward. Point-of-care ultrasound can measure the diameter of this sheath right at the bedside, through closed eyelids, in a matter of minutes, with no radiation and no need for a trip to the scanner.

A systematic review and meta-analysis of this technique found pooled sensitivity of about 90% and specificity of about 85% for detecting raised intracranial pressure, with strong overall diagnostic accuracy.⁸ An earlier meta-analysis comparing optic nerve sheath ultrasound against CT-based signs of high pressure reported even higher numbers, with sensitivity around 96% and specificity around 92%.⁹

The challenge is that this method tells you whether pressure is elevated, not exactly what the number is. It is a screening tool, not a replacement for continuous invasive monitoring in someone with a severe brain injury. The cutoff values also vary by population: in children, a smaller threshold is appropriate, and studies in pediatric patients have found that finding the right cutoff is trickier, with lower specificity when the threshold is set conservatively.¹⁰ Still, the technique has gained wide adoption in both emergency departments and ICUs because it is fast, repeatable, and entirely non-invasive.

Retinal Venous Pulsations

Another eye-based clue sits on the retina itself. In most healthy people, the central retinal vein visibly pulsates as blood flow rises and falls with each heartbeat. When intracranial pressure climbs, that pulsation can disappear. The loss of these spontaneous retinal venous pulsations is significantly associated with higher intracranial pressure and should raise suspicion of intracranial hypertension.¹¹

This is a subtler clinical sign than looking at optic nerve sheath diameter, and historically it has been underused. The classic eye-exam finding taught in medical training is papilledema, the swelling of the optic disc. But papilledema takes time to develop and can be absent even when pressure is dangerously high. Loss of venous pulsations appears to be a more sensitive early indicator. One clinical review concluded that venous pulsation loss could serve as a useful screening test for elevated pressure, especially in patients who do not yet show papilledema.¹² The limitation is that roughly 10 to 20% of the general population lacks visible venous pulsations at baseline, so an absent pulse does not automatically mean high pressure.

Transcranial Doppler Ultrasound

Transcranial Doppler (TCD) uses an ultrasound probe placed against the temple to measure blood flow velocity through the brain’s arteries. When pressure inside the skull rises, it squeezes those arteries, changing how blood flows through them. The shape of the velocity waveform shifts, and a calculated index called the pulsatility index increases.¹³

In patients with clinical features of intracranial hypertension, the pulsatility index shows a meaningful positive correlation with directly measured ICP, though it is not precise enough to replace invasive monitoring for moment-to-moment management.¹⁴ TCD is most useful as a bedside triage tool: if you suspect rising pressure but do not yet have an invasive monitor in place, a quick TCD exam can strengthen or weaken that suspicion.

CT Imaging Signs

A head CT scan does not measure pressure directly, but it can show anatomical signs that strongly suggest pressure is elevated. Clinicians look for things like compression of the third ventricle, narrowing of the lateral ventricles, obliteration of the normal grooves on the brain’s surface (sulci), midline shift, and herniation. In one study of moderate to severe traumatic brain injury patients, sulcal obliteration and third ventricular compression on early CT were the signs that correlated most strongly with subsequently measured intracranial pressure.¹⁵

CT is almost always available in hospitals, and most head-injured patients get one early on anyway, so these signs come “for free” as part of the initial workup. But they are qualitative, not quantitative. A compressed ventricle tells you pressure is probably high; it does not tell you whether you are at 25 mmHg or 40 mmHg. And some patients with elevated ICP have a CT that looks deceptively normal, so imaging alone cannot be trusted as the final word.

Waveform Analysis and Intracranial Compliance

When you have continuous ICP monitoring in place, the pressure trace is not a flat line. It pulses with each heartbeat, producing a characteristic waveform with three small peaks commonly labeled P1, P2, and P3. In a healthy, compliant brain, P1 (driven by arterial pulsation) is the tallest. As the brain loses compliance and pressure builds, P2 grows taller than P1. The ratio of P2 to P1 has been studied as an indicator of how much reserve the brain has left before small volume additions cause large pressure spikes.¹⁶

This matters because the absolute ICP number does not tell the whole story. Two patients at 18 mmHg can be in very different situations: one may have a brain that can still absorb minor changes, while the other is teetering on the edge where any small increase could cause a dangerous spike. Waveform analysis gives clinicians a window into that second layer of information.

Risks of Invasive Monitoring

Placing a catheter inside the skull is not without consequences. Infection is the most closely watched complication with EVDs. One study found the infection rate within 30 days of placement was about 2.2%, with the most common culprit being a type of skin bacteria. Patients who developed EVD infections stayed in the hospital significantly longer, averaging 23 days after placement compared to 16 days for those who did not get infected. Risk factors that stood out included prior brain surgery involving CSF diversion, CSF leaking around the catheter, and breakdown of the insertion site.¹⁷ Interestingly, simply having the drain in place for more than seven days did not independently raise infection risk in that cohort.

Bleeding along the catheter track is another concern. About 14% of EVD placements in one retrospective study were complicated by some degree of tract hemorrhage. Patients with lower platelet counts (below 120,000) had roughly five times the odds of bleeding, and those with existing brain hemorrhage were about three times more likely to develop this complication.¹⁸ These risks are part of why clinicians weigh the decision to place an invasive monitor carefully and why the search for reliable non-invasive alternatives continues.

Measuring ICP in Children

Pediatric ICP monitoring shares many of the same principles as adult monitoring, but the anatomy creates both challenges and opportunities. In infants whose skull bones have not yet fused, the soft fontanelle offers a natural acoustic window. A novel non-invasive device has been developed that measures ICP through the open fontanelle without penetrating the skull, and early clinical use in neonates with hydrocephalus from intraventricular hemorrhage showed that the pressure readings correlated with cranial ultrasound findings and changed appropriately after CSF removal.¹⁹

Optic nerve sheath ultrasound is also increasingly used in pediatric emergency departments and critical care settings, serving as a bedside extension of the physical exam when you need a quick, non-invasive read on whether a child’s pressure might be elevated.²⁰ The threshold diameters used in adults do not translate directly to younger children, so pediatric-specific cutoffs are still being refined.

Tympanic Membrane Pulsation

One of the more surprising approaches under investigation involves the eardrum. Intracranial pressure fluctuations are transmitted through the structures of the inner ear to the tympanic membrane, which pulsates subtly with each heartbeat. Researchers have explored whether analyzing the shape of this tiny pulsation can distinguish normal from elevated ICP. In one controlled study, the ratio of slopes in the tympanic membrane pulsation waveform was significantly different between normal and elevated pressure states, achieving 100% separation accuracy among the study subjects.²¹

This is still early-stage research with small numbers, so the perfect accuracy figure should be taken with caution. But the concept is appealing: a sensor placed in or near the ear canal could theoretically offer continuous, completely non-invasive ICP screening without needing ultrasound training or a clear view of the eye.

Telemetric and Implantable Monitors

For patients who need pressure data over months or years rather than days, repeated invasive procedures and hospital visits become unsustainable. Telemetric ICP monitors are implanted once and then transmit readings wirelessly, allowing clinicians to track pressure trends remotely. A systematic review of telemetric monitoring in hydrocephalus patients found that these devices enhance diagnostic accuracy and can reduce hospital admissions by catching problems before they become emergencies.²² The concept is not entirely new: a fully implantable radio-telemetric ICP sensor was described as far back as 1977 and used successfully for both short- and long-term monitoring.²³

Modern iterations are pushing toward smartphone integration. One prototype implantable sensor paired with a Bluetooth-enabled mobile application demonstrated accuracy within roughly 2 mmHg at clinically relevant pressure ranges (at or below 20 mmHg), and the app could detect simulated shunt malfunctions in real time.²⁴ Imagine a parent being alerted on their phone that their child’s shunt pressure is trending upward, hours before symptoms appear. That is the direction the technology is heading.

Machine Learning for Predicting Pressure Spikes

Another frontier is not measuring ICP directly but predicting what it will do next. Machine learning models trained on ICU data can ingest a patient’s recent vital signs and ICP trends, then forecast whether a dangerous pressure spike is coming. One model trained on a large cohort of neurointensive care patients achieved strong predictive performance for ICP increases above 22 mmHg, even when predicting up to 24 hours in advance.²⁵ A separate model demonstrated real-time continuous prediction of ICP signal patterns for the upcoming 10 minutes by learning from the preceding 20 minutes of data.²⁶

These tools do not eliminate the need for an actual pressure sensor — the models are trained on invasive ICP data and currently need that data stream as input. But they add a layer of anticipation that could give clinical teams precious lead time to intervene before pressure reaches dangerous levels rather than reacting after it already has.

ICP Concerns in Spaceflight

An unexpected setting where intracranial pressure measurement has become a hot research topic is space. Astronauts on long-duration missions frequently develop visual changes, optic disc swelling, and structural eye abnormalities collectively called spaceflight-associated neuro-ocular syndrome, or SANS. The leading hypothesis involves chronic shifts in fluid distribution: without gravity pulling blood and CSF toward the lower body, more fluid accumulates in the head. A systematic review concluded that SANS likely has a multifactorial origin, strongly affected by changes in ICP and CSF dynamics, along with factors like elevated carbon dioxide levels, intense exercise regimens, and altered sleep-wake cycles.²⁷

The problem is that you cannot exactly drill a hole in an astronaut’s skull on the International Space Station. Non-invasive methods like optic nerve sheath ultrasound and transcranial Doppler are being explored as tools astronauts could use on each other in orbit. Animal research supports this kind of long-term monitoring: studies in freely moving rats have shown that ICP fluctuates with circadian rhythm and changes predictably with anesthesia and activity state, providing baseline data that could help interpret remote readings from astronauts.²⁸ As missions to Mars and longer orbital stays are planned, developing a reliable, non-invasive, portable ICP monitor has moved from a clinical wish-list item to a genuine space-exploration priority.