MRI can detect a cerebrospinal fluid leak, and it is typically the first imaging test ordered when one is suspected. But the relationship between MRI and CSF leaks is more nuanced than a simple yes or no. A standard brain MRI often picks up indirect signs that spinal fluid is low, such as thickened meninges and a brain that appears to sag downward, rather than directly showing the hole where fluid is escaping. Pinpointing the actual leak site usually requires additional MRI techniques or other imaging altogether, and in a meaningful fraction of confirmed cases, the brain MRI looks completely normal.
What a Brain MRI Actually Reveals
When physicians suspect a CSF leak, a brain MRI with contrast is usually the opening move. The scan does not typically show the leak itself, which is almost always somewhere along the spine or skull base. Instead, it reveals the consequences of having too little fluid inside the skull. The skull is a closed box containing brain tissue, blood, and cerebrospinal fluid. When CSF volume drops, the other compartments compensate. Blood vessels expand, the meninges (the membranes covering the brain) thicken, and the brain can shift downward. MRI is very good at capturing these compensatory changes.
The hallmark finding is pachymeningeal enhancement: the thick outer meningeal layer lights up brightly after gadolinium contrast is injected. Other classic signs include engorgement of the venous sinuses, enlargement of the pituitary gland, subdural fluid collections, and in more severe cases, visible sagging of the brain itself.1Neurology. The Monro-Kellie hypothesis: Applications in CSF volume depletion One case report documented brain sagging dramatic enough to cause bilateral temporal lobe herniation alongside pachymeningeal enhancement.2PubMed Central. Frontotemporal brain sagging syndrome: Craniospinal hypovolemia secondary to a T6-T7 cerebrospinal fluid-venous fistula These findings are suggestive enough that many neurologists treat them as strong circumstantial evidence of a leak, even before the leak site is identified.
A research team developed a scoring system, now known as the Bern score, that assigns points to specific brain MRI features to estimate the probability of a spinal CSF leak. Three findings are weighted as major: pachymeningeal enhancement, venous sinus engorgement, and effacement (compression) of the suprasellar cistern, the fluid-filled space above the pituitary.3JAMA Neurology. Assessing Spinal Cerebrospinal Fluid Leaks in Spontaneous Intracranial Hypotension With a Scoring System Based on Brain Magnetic Resonance Imaging Findings Patients with confirmed leaks had significantly higher Bern scores than those without, averaging about 5.4 compared to 1.9.4PubMed Central. Relationship of Bern Score, Spinal Elastance, and Opening Pressure in Patients With Spontaneous Intracranial Hypotension The score gives clinicians a structured way to interpret what can otherwise be a subjective reading of the images.
Spinal MRI and Finding the Actual Leak
Knowing that fluid is low is helpful. Knowing where it is escaping is what makes targeted treatment possible. For this, spinal MRI enters the picture. A standard spinal MRI can reveal epidural fluid collections, essentially pools of CSF that have leaked out of the dural sac and settled in the space surrounding it. These collections are a strong clue that a leak exists somewhere nearby, even when the precise tear in the dura is not directly visible.
Heavily T2-weighted fat-saturated MRI sequences have proven particularly useful for spotting these epidural collections. This technique makes CSF appear very bright while suppressing the signal from surrounding fat, so leaked fluid stands out clearly. A study comparing different MRI sequences found that these heavily T2-weighted images with fat saturation provide high accuracy for detecting epidural CSF.5PubMed Central. Spine MRI in Spontaneous Intracranial Hypotension for CSF Leak Detection: Nonsuperiority of Intrathecal Gadolinium to Heavily T2-Weighted Fat-Saturated Sequences In a separate study, spinal MRI detected extradural fluid in 11 of 12 patients who had leaks confirmed by CT myelography, giving it a sensitivity above 90%.6PubMed. Sensitivity of MRI of the spine compared with CT myelography in orthostatic headache with CSF leak
The type of leak matters for what MRI can show. Systematic imaging studies have identified two broad categories of spinal CSF leaks. Type 1 leaks are ventral (front-facing) holes in the dura typically associated with degenerative disc disease, where a bony spur or disc fragment has worn through the membrane. Type 2 leaks are tears in the nerve root sleeves, the tube-like extensions of the dura where spinal nerves exit. Both types tend to produce epidural fluid collections that MRI can detect, though the specific tear itself is often too small to see directly on standard sequences.7American Journal of Neuroradiology. Spontaneous Intracranial Hypotension: A Systematic Imaging Approach for CSF Leak Localization and Management Based on MRI and Digital Subtraction Myelography
MR Myelography Techniques
When standard MRI detects signs of a leak but cannot pinpoint its exact location, MR myelography takes things a step further. This is still an MRI-based test, but it is specifically optimized to visualize CSF pathways and identify where fluid is escaping the dural sac.
Non-invasive heavily T2-weighted MR myelography (sometimes abbreviated HT2W-MRM) does not require any injection. It simply uses pulse sequences that make CSF extremely bright. In one series, this approach identified the CSF leak in about 80% of patients.8PubMed. Utility of heavily T2-weighted MR myelography as the first step in CSF leak detection and the planning of epidural blood patches A comparison between 2D and 3D versions of this technique found that leaks were more conspicuous on 3D MR myelography, making it more reliable for detection.9European Journal of Radiology Open. Comparing 2-dimensional versus 3-dimensional MR myelography for cerebrospinal fluid leak detection
The more invasive version involves injecting a gadolinium-based contrast agent directly into the spinal fluid (intrathecal gadolinium MR myelography). This lights up the CSF so vividly that even small leaks can become visible. One study found that this technique localized CSF leaks at cervical, thoracic, and lumbar levels with no adverse effects from the injection.10PubMed Central. Spinal Cerebrospinal Fluid Leakage in Spontaneous Intracranial Hypotension: An Intrathecal Gadolinium Enhanced MR-Myelography Study However, its diagnostic yield is limited. In patients where CT myelography had already failed to find the leak, intrathecal gadolinium MR myelography identified the site in only about one in five cases.11PubMed Central. The role of MR myelography with intrathecal gadolinium in localization of spinal CSF leaks in patients with spontaneous intracranial hypotension The technique is also considered off-label for the contrast agent, so it tends to be reserved for difficult cases where other imaging has come up empty.12PubMed. Diagnostic Yield of Intrathecal Gadolinium MR Myelography for CSF Leak Localization
When MRI Comes Back Normal
Here is where things get frustrating for patients. A normal brain MRI does not rule out a CSF leak. This is one of the most important points for anyone navigating the diagnostic process. Pachymeningeal enhancement and the other hallmark signs are present in many patients, but not all. A systematic review covering 529 patients with spontaneous intracranial hypotension found that half of them, 266 patients, were classified as MRI-negative, meaning their brain MRI showed no typical abnormalities despite having a confirmed or strongly suspected CSF leak.13PubMed Central. Spontaneous intracranial hypotension with negative brain MRI findings: a systematic review of diagnostic strategies and clinical outcomes
This is not a minor footnote. Roughly half of patients in that pooled data had unremarkable brain MRIs. The reasons are not fully understood. Some leaks may be slow or intermittent enough that the brain compensates without producing the dramatic shifts in blood volume and meningeal appearance that MRI picks up. In some patients, the MRI signs develop only after the leak has been present for a while, so early scans can miss them. One case series documented patients whose meninges initially appeared normal but developed enhancement on follow-up imaging, and others whose enhancement disappeared even though the leak continued.14PubMed. Absent pachymeningeal gadolinium enhancement on cranial MRI despite symptomatic CSF leak The findings can wax and wane over time, making a single snapshot potentially misleading.
For patients with a normal brain MRI who still have classic symptoms like positional headaches that worsen when upright and improve when lying down, the workup should not stop. Spinal MRI, CT myelography, and sometimes radionuclide cisternography can pick up leaks that the brain MRI missed.15Cephalalgia. Spontaneous spinal cerebrospinal fluid leaks
CSF-Venous Fistulas Are Especially Elusive
A relatively recently recognized type of CSF leak poses a particular challenge for MRI. CSF-venous fistulas are abnormal connections where cerebrospinal fluid drains directly into a vein rather than pooling in the epidural space. Because the fluid immediately enters the bloodstream rather than collecting outside the dural sac, these fistulas do not produce the telltale epidural fluid collections that spinal MRI relies on to detect conventional leaks. The imaging signs can be so subtle that conventional anatomic imaging often misses them entirely.16PubMed. CSF-Venous Fistulas: Anatomy and Diagnostic Imaging
Detecting these fistulas typically requires specialized myelographic techniques, either CT-based or catheter-based digital subtraction myelography, that can capture the moment contrast material crosses from the spinal fluid space into a vein. Standard MRI and even MR myelography often fall short here. This is a growing area of research, as CSF-venous fistulas may account for a substantial portion of spontaneous intracranial hypotension cases that were previously considered idiopathic.
Skull Base and Cranial CSF Leaks
Not all CSF leaks happen along the spine. Leaks through the skull base, typically presenting as clear fluid draining from the nose (CSF rhinorrhea), are a distinct clinical problem that MRI handles differently. The goal here is usually to identify a bony defect in the skull base and confirm that the draining fluid is CSF rather than nasal mucus.
MRI plays a complementary role in these cases. Fat-suppressed T2-weighted MRI detected CSF-like fluid in 90% of cases in one study, with a sensitivity around 89%.17PubMed Central. Combined HRCT and MRI in the detection of CSF rhinorrhea A specialized technique called 3D T2 DRIVE MR cisternography offers particularly clear images with effective bone and fat suppression, fast acquisition, and very bright CSF signal, all helpful for localizing exactly where fluid is escaping through the skull base.18PubMed Central. Imaging review of cerebrospinal fluid leaks
A systematic review of diagnostic methods for CSF rhinorrhea found that standard MR cisternography achieved a median sensitivity of 94% and specificity of 77%. Contrast-enhanced MR cisternography performed even better, reaching a median sensitivity of 99% and specificity of 100%.19PubMed Central. Diagnosis and Localization of Cerebrospinal Fluid Rhinorrhea: A Systematic Review These numbers make MRI-based techniques among the most accurate options for cranial leaks, though in practice most clinicians combine MRI with high-resolution CT, which is better at showing the actual bony defect. The two tests together give the surgeon both the soft-tissue detail (where fluid is pooling) and the bony anatomy (where the hole is).
How MRI Stacks Up Against CT Myelography
CT myelography, which involves injecting contrast directly into the spinal fluid and then scanning with CT, has traditionally been considered the reference standard for localizing spinal CSF leaks. The natural question is whether MRI-based techniques can match it.
The short answer is that they come remarkably close. A large study comparing heavily T2-weighted MR myelography to CT myelography in patients with spontaneous intracranial hypotension found near-perfect agreement between the two. Extradural fluid collections were identified in about 48% of patients by both methods, with an overall agreement rate above 98%.20JAMA Neurology. Computed Tomography vs Heavily T2-Weighted Magnetic Resonance Myelography for the Initial Evaluation of Patients With Spontaneous Intracranial Hypotension An earlier study similarly found no significant difference in detection rates for leaks along nerve roots (84% for MR myelography versus 74% for CT myelography) or epidural collections (89% versus 79%). MR myelography actually identified more spinal levels affected than CT myelography did.21PubMed. Heavily T2-weighted MR myelography vs CT myelography in spontaneous intracranial hypotension
The practical advantage of MR myelography is that the non-contrast version requires no injection and no radiation. This makes it more comfortable, less risky, and repeatable. CT myelography still has edges in certain situations, particularly when dynamic imaging is needed to capture fast-flowing leaks or CSF-venous fistulas in real time. But for initial evaluation, many centers now use MR myelography first and escalate to CT myelography only if the MRI-based approach is inconclusive.
CSF Leaks in Children and Adolescents
Spontaneous intracranial hypotension is uncommon in children, but it does occur, and MRI plays the same central diagnostic role. In a pediatric review, 10 of 12 children who underwent contrast-enhanced brain MRI showed dural enhancement, brain sagging, or venous distention. Spinal MRI demonstrated extensive extradural CSF collections in the majority.22American Journal of Neuroradiology. Spontaneous Intracranial Hypotension in Children: A Multi-Institutional Review of Spinal CSF Leaks Localized on Advanced Myelography
A larger review of 24 pediatric and adolescent patients found that brain MRI showed typical changes in about 79% of cases. Spinal imaging identified a CSF leak in half, with meningeal diverticula (small outpouchings of the spinal membranes) seen in many of the rest.23PubMed. Spontaneous intracranial hypotension in childhood and adolescence The roughly 20% of pediatric patients with normal-appearing brain MRIs mirrors the pattern seen in adults, reinforcing that a normal scan does not exclude the diagnosis in any age group.
Diagnosing children can be trickier because they may not describe the classic positional quality of their headaches as clearly as adults do. A higher index of clinical suspicion, combined with awareness that the MRI may be misleadingly normal, is important for pediatric cases.
Using MRI to Track Treatment Response
MRI is not only useful for diagnosis. It also helps monitor whether treatment is working. The most common first-line treatment for a spinal CSF leak is an epidural blood patch, where a small amount of the patient’s own blood is injected into the epidural space to seal the leak. After the procedure, follow-up MRI can show whether the brain MRI abnormalities are resolving: meningeal enhancement fading, venous sinuses returning to normal size, brain position normalizing.
Some centers perform a post-treatment MRI within the first week to assess the immediate response, with a repeat scan at a few months to evaluate longer-term outcomes.24British Journal of Radiology. Magnetic resonance imaging predicted the therapeutic response of patients with spinal cerebrospinal fluid leakage undergoing targeted epidural blood patch This is particularly valuable for identifying patients who may need a second blood patch or a different intervention. If the brain MRI findings persist despite treatment, it suggests the leak has not fully sealed.
Deep Learning and the Future of MRI Interpretation
One of the newer research frontiers is using artificial intelligence to read brain MRIs for signs of CSF leaks. A deep learning model was developed to predict the presence of CSF-venous fistulas based on brain MRI alone, trained on the same subtle patterns that experienced neuroradiologists look for, such as dural enhancement, pituitary engorgement, and brain sagging.25PubMed Central. Identifying Patients with CSF-Venous Fistula Using Brain MRI: A Deep Learning Approach The appeal is obvious: if software can reliably flag these findings, it could help identify patients earlier, especially in settings where expert neuroradiologists are not available to review every scan. The technology is still in early stages, but it points toward a future where MRI’s sensitivity for CSF leaks improves through better interpretation rather than better hardware alone.
The Bern scoring system mentioned earlier already moves in this direction by standardizing the reading process. Combined with automated image analysis, the goal is a diagnostic pipeline where fewer leaks slip through the cracks of a “normal-looking” MRI.26PubMed Central. Negative correlation between the Bern score and opening pressure in myelography positive spontaneous intracranial hypotension For patients who have spent months or years with debilitating headaches and a string of normal-appearing scans, tools that squeeze more diagnostic information out of the same images represent real hope.