How to Measure ICP: Invasive and Non-Invasive Methods

Intracranial pressure is measured most accurately by placing a catheter or sensor directly inside the skull, with an external ventricular drain (EVD) connected to a pressure transducer remaining the gold standard. Non-invasive alternatives exist and are improving, but none yet match the precision of invasive monitoring for clinical decision-making. The gap between the two categories is narrowing, though, as ultrasound-based and optical techniques mature. Understanding how each method works, where it excels, and where it falls short matters for clinicians choosing a monitoring strategy and for patients and families trying to make sense of what is happening at the bedside.

Why Intracranial Pressure Needs Monitoring at All

The skull is essentially a rigid box. Inside it sit three components: brain tissue, blood, and cerebrospinal fluid (CSF). The classic principle describing this relationship, known as the Monro-Kellie doctrine, holds that the combined volume of those three components stays roughly constant. If one increases, another has to decrease to compensate, or pressure climbs.¹ When the brain swells after a traumatic injury, when a tumor grows, or when CSF drainage is blocked, the compensatory mechanisms eventually run out. Pressure rises, blood flow to brain tissue drops, and the result can be permanent damage or death.

Current management guidelines generally recommend intervening when ICP exceeds roughly 20 to 22 mmHg, with a target cerebral perfusion pressure of 60 to 70 mmHg.² That threshold is the trigger for treatments like draining CSF, administering osmotic agents, or even surgery. But to know when you have crossed the threshold, you need a reliable measurement, and that is where the technical challenge begins.

When Clinicians Decide to Monitor

ICP monitoring is most commonly performed in patients with severe traumatic brain injury who have a very depressed level of consciousness. There is no top-tier (Level I) evidence mandating it, but guidelines classify monitoring as a Level II recommendation for patients whose Glasgow Coma Scale score falls between 3 and 8. Even patients with moderate injuries, scoring 9 to 12, are sometimes monitored if other signs suggest rising pressure.³ Outside of trauma, monitoring is also used in conditions like subarachnoid hemorrhage, hydrocephalus, large strokes, and brain tumors that cause swelling. The decision often depends on whether the patient’s neurological exam can be followed reliably; when someone is deeply sedated or in a coma, the exam alone cannot tell you what is happening inside the skull.

Invasive Methods

Invasive monitoring remains the only way to get a continuously accurate ICP number. All invasive approaches share the same basic idea: place a pressure-sensing element inside the skull and read the signal it sends.

External Ventricular Drain

The EVD is a thin catheter threaded through a small hole in the skull into one of the brain’s fluid-filled ventricles, then connected to an external transducer. It is considered the reference standard for ICP measurement.⁴ Beyond reading pressure, the EVD doubles as a treatment tool: opening the drain allows CSF to flow out, directly lowering ICP when it spikes.⁵ That dual function is a major advantage over other sensors. The downside is that placement requires navigating the catheter into a small target, and in patients with compressed or shifted ventricles, the procedure can be difficult or even fail entirely.⁶

Intraparenchymal Monitors

These are tiny pressure-sensing probes placed directly into the brain tissue itself, rather than into the ventricles. They are easier to insert and generally just as accurate as EVDs for reading pressure. Microtransducer-based parenchymal monitors have become a popular alternative, particularly when ventricle access is tricky.⁷ One disadvantage is that some of these devices can develop “zero drift,” a gradual shift in the baseline reading over days, which can make the numbers less trustworthy the longer the monitor stays in. Unlike EVDs, parenchymal probes cannot drain fluid, so they are purely diagnostic.

Other Invasive Sites

Lumbar drains measure pressure in the spinal CSF space rather than inside the skull. They are sometimes used in patients with communicating hydrocephalus or after certain surgeries but are less suitable when there is a risk of brain herniation. Subarachnoid bolts, screwed into the skull to sit just beneath the bone, were once more common but have largely been replaced by the devices above. Subdural and epidural sensors also exist but are used less frequently because their readings tend to be less accurate than ventricular or parenchymal measurements.

Complications of Putting Hardware in the Brain

Any device that penetrates the skull carries risks. Infection is the most frequent concern, and the longer a catheter stays in, the higher the chance of it occurring. In one comparison study, only one infection was reported among 123 patients with parenchymal monitors, versus 16 infections among 173 patients with EVDs, with catheter duration strongly linked to infection rate.⁸ Hemorrhage during placement is another risk, along with the possibility of neurological damage from the catheter itself. These complications are individually uncommon, but they add up across a large trauma population, which is one of the main motivations for developing non-invasive alternatives.

Non-Invasive Measurement Through the Eyes

The optic nerve is essentially an extension of the brain, wrapped in a sheath that communicates with the same CSF pool inside the skull. When ICP rises, that sheath expands. Measuring the diameter of the optic nerve sheath (ONSD) with a simple bedside ultrasound probe placed over the closed eyelid has become the most widely studied non-invasive ICP technique.⁹

A large meta-analysis found that ultrasound ONSD had a sensitivity of about 90% and a specificity of about 88% for detecting elevated ICP.¹⁰ Those are strong numbers for a bedside screening test. Cutoff values in the literature range from about 5 to 6 mm, with most studies settling somewhere around 5 to 5.9 mm as the threshold suggesting raised pressure.¹¹ Correlation with invasive ICP has varied across studies, with correlation coefficients reported between about 0.64 and 0.90. In children, the technique has also proven reliable, showing a good relationship with invasive ICP readings and high diagnostic accuracy for detecting raised pressure.¹²

Other eye-related approaches are being explored. Retinal blood vessels may reflect ICP changes; one study of patients with idiopathic intracranial hypertension found that the complexity of retinal venules imaged by optical coherence tomography angiography appeared to track with ICP levels.¹³ Fundoscopic examination for papilledema (swelling of the optic disc) has been used for over a century as a clinical clue to raised ICP, but it develops slowly and is a late sign rather than a real-time monitor.

Transcranial Doppler Ultrasonography

Transcranial Doppler (TCD) uses ultrasound aimed at blood vessels through the temporal bone to measure the speed and pattern of blood flowing through the brain’s arteries. When ICP rises, it compresses blood vessels and alters the flow pattern in a characteristic way. A widely used metric called the pulsatility index (PI) captures the difference between peak and trough flow velocities. One study found a strong positive correlation between PI and invasive ICP, with a correlation coefficient of 0.938.¹⁴

The catch is that PI is not specific to ICP. A rising PI can also reflect a drop in blood pressure, changes in heart rate, or shifts in the stiffness of blood vessel walls. In other words, PI goes up whenever the brain’s blood flow dynamics change, not only when pressure inside the skull goes up.¹⁵ This means TCD works best as a trend monitor or screening tool rather than a standalone replacement for direct measurement. It is also operator-dependent: getting a good ultrasound window through the skull takes practice, and some patients have temporal bone that is simply too thick.

Neuroimaging Signs of Raised Pressure

CT and MRI scans do not measure ICP directly, but they can show the structural consequences of elevated pressure. On a CT scan, features like sulcal obliteration (where the brain swells until the grooves on its surface disappear), compression of the third ventricle, midline shift, and herniation all correlate with elevated ICP. A study of patients with moderate to severe traumatic brain injury found that sulcal obliteration and third ventricular compression were significantly associated with measured ICP at or above 20 mmHg, and these imaging signs also correlated with patient outcomes at six months.¹⁶

MRI can reveal subtler signs, especially in conditions like idiopathic intracranial hypertension (IIH). These include an empty sella turcica (a flattened pituitary gland), optic nerve tortuosity, distension of the optic nerve sheath, flattening of the back of the eyeball, and stenosis of the brain’s venous sinuses. A meta-analysis found that MRI signs tend to have very high specificity, ranging from 90% to 99%, meaning they rarely give false positives. But sensitivity is generally lower, meaning a normal-looking MRI does not rule out raised ICP. The exception is sinus stenosis, which showed both high sensitivity (about 90%) and high specificity (about 96%).¹⁰ On the rarer end, the “white cerebellum sign” on CT, where the cerebellum appears brighter than the cerebral hemispheres, has been described in cases of severely elevated pressure, particularly in children with bacterial meningitis.¹⁷

Imaging is most useful for establishing whether ICP is likely elevated and identifying the cause, but it gives you a snapshot rather than continuous monitoring. A patient’s CT may look concerning, prompting clinicians to place an invasive monitor for ongoing surveillance.

Acoustic Methods and the Ear

A more unusual approach exploits the fact that the inner ear is connected to CSF through a tiny channel. Changes in ICP alter how the eardrum moves in response to sound. Tympanic membrane displacement (TMD) measurement sends a brief sound pulse into the ear and records the tiny resulting eardrum movement. In children with hydrocephalus and confirmed raised ICP, TMD readings were significantly more negative (the eardrum displaced inward more) compared to children with normal ICP.¹⁸ Reference values for healthy adults have been established, showing a wide normal range that depends on body position and which ear is tested.¹⁹

The limitation is significant: the technique requires an intact acoustic reflex pathway and a working stapedial reflex, which not everyone has. Middle-ear disease, prior ear surgery, or simply anatomic variation can make the measurement impossible. It remains largely a research tool rather than something you would find in routine clinical use.

Near-Infrared Optical Sensing

One of the newer frontiers uses near-infrared light shone through the skull. The principle is that changes in ICP alter the pulsatile characteristics of blood flow in the brain, and those alterations subtly change how near-infrared light is absorbed and scattered. A study in healthy volunteers showed that features extracted from near-infrared spectroscopic signals could identify ICP changes induced by position changes and straining (Valsalva maneuvers), with sensitivity reaching up to 80% and specificity around 70%.²⁰ A separate clinical study applied similar optical sensors alongside invasive ICP monitors in 27 patients with traumatic brain injury, testing whether machine-learning algorithms could estimate continuous ICP values from the optical signals.²¹

This technology is still early-stage. The accuracy is not yet reliable enough for clinical decisions, but it has the appeal of being completely non-invasive, portable, and potentially wearable. If the signal-processing improves, optical sensors could eventually serve as continuous monitors in settings where drilling a hole in the skull is impractical.

How Non-Invasive Methods Stack Up Overall

The honest summary is that no non-invasive technique is reliable enough to replace invasive monitoring when accurate, continuous ICP numbers are needed to guide treatment.⁷ Ultrasound ONSD comes closest to a clinical workhorse: it is fast, repeatable, and has reasonably high diagnostic accuracy for detecting whether ICP is elevated above a dangerous threshold. But it gives you a binary answer (likely elevated or likely normal) more than a precise number, and it cannot provide the minute-to-minute trend data that guides treatment in an intensive care unit. TCD, imaging, TMD, and optical methods each have their niche, but all share some combination of limited specificity, dependence on operator skill, or insufficient accuracy at the individual-patient level.

Where non-invasive tools shine is in screening and triage: deciding who needs the invasive monitor in the first place, following trends in patients who are not sick enough to justify a catheter, and monitoring in settings where surgery is not available.

Monitoring in Infants

Babies present a unique situation because their skull bones have not yet fused. The anterior fontanelle, the soft spot on top of the head, transmits ICP changes directly to the skin surface. Clinicians have traditionally assessed infant ICP by palpating the fontanelle for bulging or measuring head circumference. These methods are crude but cost nothing and require no equipment. A fontanometer, a device designed to press gently against the fontanelle and measure the tension, has been tested as a way to move beyond subjective palpation. One proof-of-concept study noted that while invasive monitoring is the standard for accuracy, its need for surgery makes it impractical for routine use in neonates, and existing non-invasive methods are generally not designed for this age group.²² The fontanelle window closes by about 18 months of age, so this approach has a narrow but important time frame.

Emerging Wireless and Implantable Devices

Several research groups are working to bridge the gap between invasive accuracy and non-invasive convenience. One prototype is an implantable, rechargeable ICP monitor that communicates via Bluetooth to a smartphone app. In bench testing, it met accuracy requirements at clinically relevant pressures (error within about 2 mmHg at pressures at or below 20 mmHg) and successfully simulated the detection of shunt malfunctions and abnormal ICP wave patterns.²³ The concept is aimed primarily at patients with shunts for hydrocephalus, who currently must come to the hospital whenever shunt malfunction is suspected. A wireless implanted sensor could let them check at home.

Another approach uses ultrathin piezoelectric resonators that can be implanted under the skull and interrogated wirelessly without a battery. One such device demonstrated a detection limit of 0.15 mmHg across a range up to 240 mmHg in laboratory testing.²⁴ These are still lab prototypes, not devices you would find in a hospital today, but they represent the direction the field is heading: accurate, continuous, and minimally burdensome for the patient.

ICP Monitoring in Space

An unexpected frontier for ICP measurement is spaceflight. Astronauts on long-duration missions develop a constellation of eye changes known as spaceflight-associated neuro-ocular syndrome (SANS), which may involve fluid shifts toward the head, inflammation around the optic nerve sheath, and changes in CSF dynamics.²⁵ NASA has used ultrasound ONSD measurements aboard the International Space Station as a screening tool. MRI scans of 27 astronauts taken about 100 days after flight showed a mean ONSD of 6.2 mm, above the thresholds commonly associated with intracranial hypertension on the ground.²⁶

Placing an EVD in orbit is obviously impractical, so the push for non-invasive methods is existential for long-duration spaceflight medicine. The same ONSD ultrasound that serves as a useful but imperfect screening tool on Earth becomes the best available option when you are hundreds of miles above it. As missions to the Moon and Mars grow closer, developing portable, non-invasive ICP monitors that work reliably in microgravity is a specific engineering and medical priority for space agencies.

Waveform Analysis and Precision Monitoring

Regardless of which invasive device is used, all of them generate an ICP waveform, a pulsating signal that rises and falls with each heartbeat. The simplest use of this waveform is to read the mean pressure and compare it against the treatment threshold. But the waveform contains more information than a single number. Its shape, specifically the relationship between the first and second peaks (often labeled P1 and P2) and the time it takes to reach its peak, can indicate how well the brain is coping with pressure changes. When P2 is taller than P1, intracranial compliance is reduced, meaning the brain is running out of room to compensate, sometimes even before the mean ICP number looks alarming.²⁷

Some newer non-invasive systems, like strain-gauge sensors placed against the skull, attempt to detect this waveform through intact bone and use the P2/P1 ratio as a surrogate for invasive monitoring. Early results have been mixed. Sensitivities of 70% to 100% have been reported, but specificities have ranged as low as 41%, meaning a fair number of false alarms.¹¹ Getting a clean waveform through the skull is technically demanding, and signal quality varies a lot between patients. The concept is promising because it targets the information clinicians find most useful, but the technology is not yet consistent enough for widespread adoption.