Important Note:

First and foremost, if your child has developmental delays or other concerns, it is always advisable to consider an EEG, as it can provide valuable insights.

EEG Abnormalities in Awake and Sleep States – What Does It Mean?

An EEG measures the brain’s electrical activity. Abnormalities on EEG (especially patterns called epileptiform discharges, such as spikes or spike-and-wave complexes) indicate a tendency for seizures or a disturbed brain function often seen in epilepsy. When a child’s EEG is abnormal both when awake and asleep, it usually means the brain has ongoing epileptic activity even outside of obvious seizures. In young children, especially those with developmental or language regression, this often points to an epileptic encephalopathy – a condition where the epileptic activity itself contributes to cognitive and language impairments.

Importantly, EEG abnormalities can be much more pronounced during sleep in some syndromes. During non-REM sleep, the brain’s networks are more synchronized, and epileptic spikes can become nearly continuous in certain conditions. This is why doctors often do overnight or sleep EEGs if a daytime EEG is inconclusive. If the EEG during sleep shows near-constant spikes, it could indicate a syndrome called ESES (Electrical Status Epilepticus in Sleep). In ESES, the child’s EEG shows continuous or near-continuous spike-and-wave discharges during slow-wave sleep, often exceeding 85% of the sleep time. Such intense nighttime EEG activity can disrupt the brain’s ability to consolidate learning and language, leading to regression or stagnation in development.

In summary: EEG abnormalities in both awake and sleep states suggest an ongoing epileptic disturbance, which in a young child correlates with conditions like LKS/ESES or other forms of epilepsy that can significantly affect development. It is a red flag that the child’s brain is under almost constant “electrical stress,” which can manifest as language and cognitive difficulties.

Landau–Kleffner Syndrome (LKS) – The “Epileptic Aphasia”

One possibility for a previously healthy (or minimally delayed) child who becomes nonverbal and cannot understand speech is Landau–Kleffner Syndrome (LKS). LKS is an extremely rare epilepsy syndrome typically presenting between 3 and 6 years of age. It is characterized by acquired aphasia (loss of language abilities) and epileptic EEG abnormalities. In LKS, children usually have normal early development, then lose language comprehension and speaking ability over time. A classic sign is verbal auditory agnosia – the child can no longer understand spoken language or even recognize familiar sounds (they may not distinguish a dog’s bark or the doorbell). Parents often notice the child seems deaf or ignores verbal cues; indeed, LKS children are sometimes misdiagnosed as having hearing loss or autism before the true cause is found. Behavioral problems like hyperactivity or attention deficits can accompany LKS.

EEG findings in LKS: An EEG in LKS typically shows epileptiform discharges in the language areas of the brain, often over the temporal lobes (around the Sylvian fissure), especially during sleep. In fact, the hallmark is that EEG abnormalities dramatically worsen during slow-wave sleep, sometimes meeting the definition of ESES (continuous spike-waves in sleep). During the day, the EEG might show only infrequent spikes or be less abnormal, but at night it “explodes” into near-constant epileptic activity. This pattern disrupts the child’s ability to process language and learn during critical developmental years.

It’s important to note that not all children with LKS have obvious motor seizures. Some have “subclinical” seizures or only EEG seizures that manifest primarily as language and cognitive regression. In some cases, parents might not witness convulsive seizures at all – the EEG changes and language loss are the main clues. This is why LKS is considered an epileptic encephalopathy (the epileptic activity itself causes the encephalopathic features like loss of skills).

Diagnosis: LKS is diagnosed by the characteristic clinical picture (acquired language loss in a 3-6 year old) combined with an EEG that shows epileptic discharges, especially in sleep, often maximal in one hemisphere (commonly the left side which controls language in most children). A sleep EEG is crucial; a routine short EEG might miss the diagnosis. As the Texas Children’s Hospital notes, an EEG done over a longer period during sleep is most effective at capturing the abnormalities in LKS. Neuroimaging (MRI) is usually normal in LKS – there is no structural brain lesion to see – but MRI may be done to rule out other causes. It’s also wise to perform hearing tests to ensure the issue is truly in the brain’s language processing, not the ears.

Is LKS the same as ESES or autism?

• LKS vs ESES/CSWS: LKS is closely related to the EEG phenomenon of ESES. In fact, LKS can be considered one end of a spectrum of epileptic encephalopathies with continuous spikes during sleep. The defining feature of LKS is the language regression (aphasia). Not all children with the ESES EEG pattern have the selective language loss; if they have more global cognitive regression, it might be labeled differently (such as CSWS – Continuous Spike-Waves in Sleep syndrome). But essentially, LKS is a specific clinical syndrome associated with ESES, where the impact is mainly on language.
• LKS vs Autism: LKS can superficially resemble autism or an autistic regression, because the child stops responding to language and may withdraw socially. However, there are critical differences. LKS involves an electroencephalographic abnormality (epileptic brain activity), whereas autism spectrum disorder (ASD) is defined behaviorally and does not inherently feature epileptic EEG patterns (although some children with ASD do coincidentally have EEG spikes). In LKS, the child’s loss of language is due to previously acquired language skills being destroyed by epileptic activity, whereas in autism, language development is usually intrinsically impaired or regresses for unclear reasons without epileptic activity. Moreover, an MRI in LKS is typically normal, whereas in autism it’s also usually normal – so imaging doesn’t distinguish them, but the EEG does. If an EEG is clearly abnormal in the pattern we described, and there’s language loss, LKS is a strong consideration rather than primary autism. (One should also consider an auditory processing disorder or hearing loss, but those won’t cause an abnormal EEG).

Treatment of LKS: Early and aggressive treatment is important. Standard anti-seizure medications (anticonvulsants) are often tried first and can sometimes control the seizures/spikes. Some cases respond to high-dose steroids or adrenocorticotropic hormone (ACTH) injections (borrowing a strategy from other epileptic encephalopathies like infantile spasms). The ketogenic diet has also shown benefit for some children with LKS. If medications fail, an option is a surgical procedure called multiple subpial transection, which involves making tiny cuts in the speech area cortex to prevent seizures from disrupting language – this can sometimes improve language if the epileptic focus is localized, but it’s a specialized procedure only for select cases. Throughout treatment, supportive therapies (speech therapy, cognitive and behavioral interventions) are essential to help the child regain or maintain communication skills.

Bottom line: LKS is a rare but treatable cause of severe language regression. The EEG (especially sleep EEG) is key to diagnosis. If LKS/ESES is confirmed, treating the epilepsy can sometimes allow language to improve, whereas purely autistic regression would not respond to epilepsy treatments. Thus, identifying LKS versus autism has important implications for therapy.

Electrical Status Epilepticus in Sleep (ESES/CSWS) – A Broader Epileptic Encephalopathy

Closely related to LKS is the broader phenomenon of ESES, also known as Continuous Spike-Wave in Slow-Wave Sleep (CSWS). This is an EEG pattern and syndrome where any sleep (especially slow-wave sleep) is occupied by nearly continuous epileptic discharges. Children with ESES/CSWS typically present between ages ~4 to 7 (slightly later median than LKS). The hallmark is neurocognitive or developmental regression (loss of skills or slowing of intellectual development) along with seizures that occur predominantly during sleep. Unlike LKS which mainly impacts language, CSWS can affect multiple domains – language, cognition, behavior, even motor skills depending on which brain regions are most affected by the nocturnal epileptic activity.

Symptoms and EEG: In CSWS/ESES, the most common seizure type is focal motor seizures during sleep (for example, partial seizures causing jerking of one side). Sometimes these seizures are subtle or not noticed at night. But the child may show daytime consequences: memory problems, hyperactivity or behavioral changes, or loss of milestones they previously mastered (e.g., worsening cognitive test scores or behavior). By definition, the EEG during slow-wave sleep shows spikes and spike-waves continuously or almost continuously (typically a spike-wave index ≥85%, meaning 85% of slow-wave sleep is occupied by epileptiform activity). Often these discharges can be generalized or multifocal, sometimes with a predominance in one region (for example, a focus in the left hemisphere could drive a hemi-ESES, where that side of the brain shows continuous spikes). If a child’s EEG report mentions “spikes in the left cortex during sleep” or “left temporal spikes with ESES,” it means the left side of the brain (often language area) is the primary generator of the epileptic activity.

Could it be FCD causing ESES? Yes, it’s possible. ESES/CSWS can be idiopathic (no clear structural cause) or symptomatic. In many children, no lesion is found (and they might label it LKS if language is the main issue). However, some known causes of ESES/CSWS include brain malformations, genetic or metabolic conditions. For example, a child with a focal cortical dysplasia or an old stroke or a polymicrogyria in one area of the brain might develop CSWS – the structural lesion in the brain serves as a nidus for continuous discharges during sleep. So, if the EEG shows a left cortical focus, doctors will often do an MRI to see if a lesion in the left cortex (like an FCD or other malformation) is present. Focal Cortical Dysplasia in the left hemisphere, for instance, could present as a child with seizures (maybe subtle) and developmental delay, and the EEG might show continuous left-sided spikes in sleep. In that case, treating or removing the FCD (if possible) might cure the epilepsy (more on FCD below).

Diagnosis and workup: Similar to LKS, diagnosing CSWS involves capturing the characteristic EEG pattern during sleep. Epilepsy clinics often perform overnight video-EEG monitoring for children with developmental regression and suspected CSWS, to quantify how much of the sleep is occupied by spikes. On the clinical side, a detailed neurodevelopmental assessment is done to document what skills have regressed. An MRI is usually ordered to look for any structural brain abnormality (malformation, injury, etc.) that might be causing the epilepsy. Blood tests for metabolic or genetic causes may be considered if no obvious structural cause is found, since conditions like metabolic disorders can sometimes trigger epileptic encephalopathies.

Treatment: The first-line treatment is anti-epileptic medication aimed at reducing the spike activity. Certain medications (like high-dose benzodiazepines at night, e.g. diazepam or clobazam, or steroids like prednisolone or ACTH) have been reported to suppress the continuous spikes and improve outcomes in ESES/CSWS, even though these may not be standard anti-seizure meds for other epilepsies. The Texas Children’s Epilepsy Center notes that early intervention is key, and while various antiseizure medications may help, in some cases surgery is considered. Surgery for CSWS might be an option if there is a clear focal lesion or one hemisphere causing the problem. For example, if an MRI finds a dysplasia or if one hemisphere is overwhelmingly abnormal, a focal resection or functional hemispherectomy might be curative. Another surgical approach for ESES (without a visible lesion) is, like in LKS, multiple subpial transections, which can interrupt the spread of epileptic activity without removing brain tissue. However, surgery is not always an option – it depends on finding a resectable focus. Many children with CSWS are managed medically, and the condition may gradually improve as they get older (ESES is often described as “self-limited” in that it eventually resolves in adolescence, but not without potential lasting developmental impacts if not treated).

Prognosis: If recognized early and treated effectively, some children can regain skills. Others may continue to have learning disabilities or require special education due to the period of developmental disruption. Every child is different – some have mild CSWS and minimal regression, others have severe regression (e.g., going from speaking to nonverbal, or losing motor skills).

In summary, ESES/CSWS is on the same spectrum as LKS but with potentially broader effects than just language. It’s crucial to investigate any child with “autism-like” regression for possible ESES, because unlike idiopathic autism, this is caused by epilepsy and often can be at least partially treated. The presence of EEG abnormalities is the key differentiator.

Focal Cortical Dysplasia (FCD) – Could a Hidden Brain Malformation be the Culprit?

Another major consideration in a young child with seizures or EEG abnormalities and developmental delay is a structural abnormality in the brain – and one of the most common in children is Focal Cortical Dysplasia (FCD). FCD is a malformation of cortical development, essentially a small area of the brain’s cortex that formed abnormally before birth. This patch of cortex has disorganized layers or abnormal cell types, and it is highly epileptogenic (prone to causing seizures). In fact, FCD is one of the most common causes of drug-resistant epilepsy in children.

How FCD presents: If an FCD is severe and in a critical area, seizures may start early in life (even infancy) and can cause developmental problems. For example: In a 3-year-old, an FCD in the left cortical region (especially left temporal or perisylvian region) could manifest as epileptic discharges on EEG coming from the left cortex, which might line up with language difficulties (since left side controls language in most right-handed individuals). The child might have had subtle seizures that went unrecognized (like brief staring episodes or nocturnal seizures), or only the EEG hints at ongoing epileptic activity. Over time, uncontrolled seizures or continuous spikes from an FCD can lead to global developmental delays or autism-like features. Unlike LKS (which has no lesion), FCD is an actual physical lesion in the brain – albeit often a very small or subtle one.

EEG clues: FCD typically causes focal epileptiform abnormalities – spikes or sharp waves in a certain region – often consistently coming from the same area (because it’s a fixed lesion). In this case, “abnormalities in the left cortical region” on EEG (especially if consistently in the same location) could point toward an FCD in that region. During sleep, epileptic abnormalities from FCD often increase (sleep tends to activate interictal spikes), so seeing more left cortical spikes in sleep is common in FCD as well. However, just an EEG can’t diagnose FCD; it only tells us location. We need imaging for that.

MRI and the challenge of “hidden” FCD: The MRI scan is the primary tool to detect FCD and other malformations. Typical MRI signs of FCD include: cortical thickening, blurring of the gray-white matter junction, abnormal signals in the white matter (especially a distinctive transmantle sign – a streak of signal from the cortex toward the ventricle) and abnormal folding patterns. FCD Type IIb, in particular, often shows the transmantle sign (seen in ~94% of cases) making it easier to spot on MRI. On the other hand, FCD Type I (a more subtle form) may look almost normal on MRI or have only very slight blurring of the cortex, often in the temporal lobe.

There are different types of FCD according to the International League Against Epilepsy (ILAE) classification:

• Type I: mild abnormalities of cortical layering (subtle microdysgenesis). MRI is often normal or very subtle in these cases. These might present later in childhood or even adulthood because they are “milder” dysplasia pathologically (though they can still cause seizures).
• Type II: more severe dysplasia with abnormal cells (Type IIa with dysmorphic neurons, IIb with additional balloon cells). These are usually more epileptogenic early in life. MRI often can detect Type II, especially IIb, because they tend to cause more obvious changes – e.g., cortical thickening, a transmantle band of abnormal signal.
• Type III: a dysplasia found in association with another lesion (for example, Type IIIa is an FCD next to a hippocampal sclerosis scar in temporal lobe epilepsy, IIIb is an FCD adjacent to a tumor like a DNET, etc.). In Type III, the MRI might be dominated by the primary lesion (tumor or scar), and the dysplasia is a secondary finding.

The question “how many types of FCD and how urgent is it to find it?” – as noted, there are essentially three broad categories (I, II, III with subtypes). The urgency to find it is high if a child has intractable epilepsy or developmental regression, because if an FCD is the cause and can be localized, surgical removal of that small abnormal area can often cure the epilepsy or greatly improve development. Epilepsy surgery is not just for seizures – in young children, stopping seizures or continuous spikes can allow the brain to resume more normal development. Therefore, epilepsy centers are quite aggressive in searching for an FCD or other lesion if the EEG suggests a focal source and the child is not developing normally. They “don’t want to lose a hidden FCD” by missing it on MRI.

Unfortunately, some FCDs are truly “MRI-negative” – meaning even a high-quality MRI doesn’t obviously show the lesion. It’s estimated that a fair number of FCDs (especially Type I or small Type II) may be missed on standard MRI. Epilepsy specialists address this by:

• Using special MRI protocols (thin slices, specific angles, sometimes 3T or even 7T MRI for better resolution).
• Using MRI post-processing techniques or computer algorithms that enhance subtle cortical features. These can sometimes flag areas of slight blurring or thickness that radiologists might overlook.
• Incorporating other imaging modalities: PET scans (positron emission tomography) can show areas of decreased metabolism in the brain, which often correspond to the epileptic focus – an FCD usually appears as a focal area of hypometabolism on interictal PET. So if MRI is normal but PET shows one temporal lobe uses less glucose, that’s a clue.
• MEG (Magnetoencephalography) can localize interictal spikes via magnetic fields and may point to a focal source even if MRI is normal.
• High-density EEG or invasive EEG (stereo-EEG): In tough cases, they might do EEG with electrodes placed inside the skull to hone in on a small abnormal area.

So, epilepsy clinics take a multimodal approach to ensure a hidden FCD isn’t missed. The combination of advanced MRI techniques, PET, and EEG can significantly improve chances of finding an “MRI-negative” FCD.

What if the EEG is abnormal but MRI doesn’t show anything? This scenario is common. It does not mean the child is fine; it just means no obvious lesion was seen on the scan. The possibilities in that case include: a very subtle FCD or malformation that current MRI can’t resolve, a primarily functional problem like LKS/ESES (which doesn’t have a structural lesion), or a genetic epilepsy where the whole brain is microscopically affected but not in a way MRI can see. When MRI is normal, doctors might label the epilepsy as “cryptogenic” or “unknown cause” initially, but they will keep looking if clinically warranted. In practice, if MRI is normal and suspicion for FCD is high (based on EEG focusing on one spot), the team might get a second neuroradiologist to review the MRI, possibly do a repeat MRI on a better machine or with special sequences, or go to PET/MEG as mentioned. It’s also worth exploring genetic testing at that point – many early childhood onset epileptic encephalopathies (Dravet syndrome, SCN2A-related disorders, etc.) can present with developmental delay and abnormal EEG, and these would have normal MRI. A genetic panel or even whole exome sequencing might reveal a cause if imaging does not.

To answer directly: If an EEG shows left cortical abnormalities but the MRI is normal, the possibilities include:

• A microscopic focal dysplasia that is hard to see (MRI-negative FCD).
• An epileptic syndrome like LKS/ESES where MRI is expected to be normal (functional disorder).
• Other malformations that are subtle or in areas hard to image.
• A genetic epilepsy (no lesion, but a channelopathy or metabolic issue causing seizures and delay).

In all cases, further workup is needed because an abnormal EEG in this context is not “normal” for autism or developmental delay – it points to something neurological we might be able to treat.

Other Conditions Mimicking Autism with EEG Abnormalities

Aside from LKS/ESES and FCD, what else could cause a nonverbal 3-year-old with abnormal EEG? There are several conditions to consider, spanning structural, genetic, and metabolic categories. Some notable ones include:

• Lennox–Gastaut Syndrome (LGS): This is a severe epilepsy syndrome that typically starts around ages 3-5. LGS is characterized by multiple seizure types (especially drop seizures, tonic seizures, atypical absences) and a characteristic slow spike-and-wave EEG pattern (around 1.5–2.5 Hz spike-wave), often with fast rhythms during sleep. Children with LGS have cognitive impairments and often developmental regression. Many LGS patients have an underlying brain abnormality or injury, but some do not. Behaviorally, children with LGS can have autistic features or other neurobehavioral problems (hyperactivity, aggression). If for example, our 3-year-old patient had not just language issues but also frequent seizures of multiple types, LGS could be on the table. MRI in LGS might show an underlying cause (e.g., prior brain injury, malformation) in some cases, or it could be normal (then it’s called cryptogenic LGS). The key difference is LGS usually involves more obvious seizures and a different EEG signature (slow generalized spike-wave, not just during sleep but also in wake). So, if the EEG abnormality was only during sleep, LGS is less likely; LGS has awake EEG abnormalities too (but slower frequency). Still, it’s another epileptic encephalopathy that causes global delay. Treatment for LGS is often challenging and may include broad-spectrum antiseizure medications, ketogenic diet, or devices like vagus nerve stimulators.
• Tuberous Sclerosis Complex (TSC): This is a genetic disorder where benign tumors (tubers) form in the brain and other organs. TSC is a major cause of early epilepsy and developmental delay, and it has a high association with autism spectrum disorder (up to ~50-60% of TSC patients have ASD or autistic traits). If a child has tuberous sclerosis, the MRI will usually clearly show cortical tubers – they are typically triangular lesions in the cortex that extend toward the ventricles, best seen on MRI T2/FLAIR images. These tubers act like FCDs (in fact, a tuber is essentially a form of cortical dysplasia caused by a TSC gene mutation). They cause seizures (infantile spasms often, and later other seizures), and can result in cognitive impairment and autism-like behavior. For example, a child with TSC might be nonverbal and have epilepsy; the difference is MRI isn’t normal – it will show multiple lesions all over the brain. In our context, TSC would likely have been considered if the MRI found something, but the question suggests perhaps MRI was normal or only a subtle left focus. Still, it’s important to know that TSC is an example of “autism but not autism” – it’s a medical condition that often leads to autistic symptoms plus EEG abnormalities and is identifiable by MRI and genetic testing. If suspected, a detailed skin exam (for ash-leaf spots or facial angiofibromas), heart ultrasound (for cardiac rhabdomyomas), and genetic test can confirm TSC. Early diagnosis is crucial because there are even mechanistic treatments now (like everolimus for certain growths).
• Polymicrogyria (PMG): This is another malformation of cortical development, where the brain has too many small folds. Polymicrogyria can affect one region or many; a common subtype is bilateral perisylvian polymicrogyria, which can cause significant speech and swallowing difficulties (because the perisylvian region includes language cortex and motor areas for mouth/throat). Children with PMG often have seizures (90% will develop epilepsy) and various degrees of developmental delay. They might have speech and motor delays, cerebral palsy features, and sometimes autistic traits. MRI is diagnostic for polymicrogyria – it will show an irregular cortical pattern with too many small gyri and an abnormal gray-white junction. If the MRI shows polymicrogyria, then we have our answer for the cause of EEG abnormalities and delays. Polymicrogyria can be associated with in utero infections like CMV or certain genetic mutations. Treatment is symptomatic (seizure control, therapies) unless a large focal area could be resected (rarely curative if bilateral). Polymicrogyria is an example of a condition that mimics autism (children may have similar symptoms) but is distinguishable by MRI, and it comes with epilepsy.
• Gray Matter Heterotopia: This is a condition where some neurons are in the wrong place (for example, nodules of gray matter along the ventricles – periventricular heterotopia, or a “double cortex” band of gray matter). These too can cause seizures and developmental issues. Girls with a “double cortex” (subcortical band heterotopia, often due to a DCX gene mutation) can have severe epilepsy and intellectual disability – sometimes with autistic features – and MRI will show the extra band of cortex. Periventricular heterotopia might cause milder issues in some (often seizures in teens), so probably less relevant to a 3-year-old, but worth mentioning as part of malformations detectable on MRI.
• Genetic Epileptic Encephalopathies: There are numerous genetic disorders where children have global developmental delay and epilepsy without a visible MRI lesion. For example, Angelman Syndrome is a neurogenetic disorder (involving the UBE3A gene) in which children have severe speech impairment (often minimal or no words), ataxia, a happy demeanor, and nearly all have seizures and an abnormal EEG. Angelman’s EEG is typically very abnormal, with characteristic high-amplitude slow waves and spikes even when they’re not actively seizing. These children are often misdiagnosed as autistic early on (because of lack of speech and intellectual disability), but genetic testing confirms Angelman. Rett Syndrome (in girls) is another – normal early development followed by regression, loss of speech and hand skills, and seizures starting by age 2-3. MRI in these genetic syndromes might be unremarkable or show only non-specific changes (like mild atrophy). Metabolic disorders (like some leukodystrophies or mitochondrial disorders) can also present around age 2-3 with regression and seizures; MRI often shows white matter changes or other specific brain changes in those cases, and metabolic/genetic tests are needed.
• Autoimmune or Post-Infectious Encephalitis: Though less common in a 3-year-old, there are cases of autoimmune encephalitis (like anti-NMDA receptor encephalitis) or epilepsy triggered by immune mechanisms that can cause developmental regression or psychiatric symptoms alongside seizures. These often have MRI changes (like inflammation) or require spinal fluid tests for diagnosis. Usually they present more acutely (over days to weeks with seizures, movement disorders, etc.) so they may be less likely in a slow developmental issue scenario. Still, epileptic encephalopathy due to an unknown inflammation could be considered if there’s a fluctuating or rapid course.
• “Hearing Impairment or Other Developmental Disorders”: Strictly speaking, if a child is nonverbal and not responding to speech, one must evaluate hearing. Severe hearing loss can cause a child to appear autistic and not develop language. But hearing loss would not cause EEG epileptic abnormalities. So if the EEG is truly abnormal, we’re looking at a neurological cause, not a primary sensory deficit or purely behavioral disorder. It’s worth mentioning only to emphasize that any child with speech delay should have a hearing test, but an abnormal EEG steers the diagnosis away from simple hearing issues into the realm of epilepsy.

Diagnostic Steps: From EEG to MRI and Beyond

When faced with a child who has global developmental delay, language delay, and an abnormal EEG, epilepsy clinics follow a systematic approach:

1. Detailed Clinical Evaluation: This includes developmental history (was the child ever talking? did they lose words or never gained them?), family history of seizures or neurological disorders, pregnancy and birth history (any complications that might cause brain injury), and a thorough physical and neurological exam. Sometimes subtle clues like skin spots (e.g. ash leaf spots for TSC), abnormal head growth (microcephaly can hint at genetic syndromes like Rett or Angelman, macrocephaly might hint at metabolic or genetic causes), or asymmetry in the child’s body (could hint at a brain malformation) are sought.
2. EEG (Electroencephalogram): A standard awake EEG is often the first test if epilepsy or regression is suspected. If it shows abnormalities (like in our scenario, spikes in awake record), the next step is often to get a sleep EEG or overnight video-EEG because certain syndromes only manifest clearly in sleep. Many epilepsy centers will admit the child for a 24-hour EEG monitoring to capture a full sleep cycle. This helps identify patterns like ESES/CSWS or frequent sleep-activated spikes. The EEG also tells us if the problem is focal (coming predominantly from one area, e.g. left cortex) or generalized. This information is critical to guiding the search for causes.
3. Neuroimaging (MRI Brain): The next routine step after confirming epileptiform EEG activity is to perform a high-resolution MRI of the brain. As noted earlier, the MRI can reveal structural causes: malformations of cortical development (like FCD, polymicrogyria, heterotopias), tumors or benign lesions (like a small tumor or tuber), old stroke or injury, or neurodegenerative changes. In an epilepsy center, MRI is often done on a 3 Tesla MRI with an epilepsy protocol (thin slices through the temporal lobes, etc.) to maximize the chance of finding subtle lesions. If the first MRI is normal but suspicion remains high (for example, a very focal EEG), they might do MRI with different techniques or at a higher field strength, or send the images to specialized centers for review. MRI is essential for planning treatment – if a lesion is found, the conversation may shift to how to treat that lesion (medically vs surgically). If no lesion is found, it suggests either a microscopic lesion or a non-structural cause like LKS.
4. Laboratory Tests: Depending on the case, doctors may run metabolic tests (blood and urine for metabolic disorders that can cause seizures) or genetic tests. Today, for a child with unexplained developmental delay and epilepsy, a genetic epilepsy panel or even whole exome/genome sequencing is often recommended, especially if MRI is normal. These tests can identify syndromes such as Angelman, Rett (in girls), SCN gene mutations, etc., which might not be obvious clinically but have important implications (for example, some genetic epilepsies might respond to specific medicines or diets). If a specific epilepsy syndrome like LKS is suspected, genetic tests likely won’t reveal anything (since LKS is not yet tied to a single gene), but the testing might still be done to rule out other things.
5. Specialized Imaging (if needed): If EEG shows a focal area but MRI didn’t show an obvious lesion, the team might arrange a PET scan of the brain. PET can identify areas of hypometabolism that correlate with seizure foci – for instance, an FCD in the left temporal lobe often shows up as a “cold” spot on FDG-PET. Another tool is SPECT (Single Photon Emission CT) during a seizure, which can sometimes show an area of high blood flow where the seizure originates; this requires capturing a seizure during an inpatient stay with an injection at the moment of seizure, which is complex especially in a 3-year-old. Magnetoencephalography (MEG), available in some centers, can noninvasively map interictal spikes with good spatial precision, possibly highlighting a focal cortical area responsible. These are usually part of a presurgical evaluation if surgery is being considered.
6. Treatment Trial and Follow-up EEGs: While all the above is happening, doctors will usually not wait to start treatment. If the child has frequent spikes or seizures, they often initiate an antiseizure medication appropriate for the suspected syndrome. For example, in suspected LKS/CSWS, they might try a benzodiazepine or antiseizure drug known to help in that condition. If the child improves (e.g., starts understanding a bit more, EEG improves), that further supports the diagnosis. Repeated EEGs are done to monitor how the brain’s activity changes with treatment.
7. Multidisciplinary Interventions: Regardless of cause, the child will benefit from supportive care: speech therapy (especially if language is delayed), occupational and physical therapy if motor skills are affected, behavioral therapy if there are ADHD or autistic-like behaviors, and special education services. These therapies should run in parallel with medical investigations and treatments. For instance, a child with LKS will need speech therapy to regain language once the epileptic activity is controlled. A child with global delay may need early intervention programs even while we are searching for the cause.
8. Second Opinions / Epilepsy Center Referral: If initial workups do not yield answers or seizures are hard to control, referral to a specialized pediatric epilepsy center is common. There, a team of pediatric epileptologists, neurosurgeons, neuropsychologists, neuroradiologists, and others will review the case in an epilepsy surgery conference (even if surgery is just a consideration). The goal is to ensure nothing is missed: Was that MRI really normal or is there a subtle FCD we can see on re-review? Do we need to do invasive monitoring to pinpoint a focus? The mention of “not losing a hidden FCD” reflects this – experts will meticulously scour the data to find any treatable lesion. In some cases, they might employ advanced MRI post-processing algorithms that statistically compare the child’s brain structure to a normal database to find subtle cortical abnormalities.

Note: The “norm” in hospitals for global delay + language delay + abnormal EEG is:
EEG -> MRI -> further investigations (genetic/metabolic tests, prolonged video-EEG, PET, etc.). Each step informs the next. This workup ensures that if it’s something obvious like a brain malformation, we find it; if it’s something elusive like LKS, we diagnose it by EEG; if it’s something unexpected like a genetic syndrome, we confirm it; and importantly, we start treatments early to give the child the best chance at improving.

MRI Findings in Focal Cortical Dysplasia (FCD) and EEG Spikes, and the Role of AI in Epilepsy Care

Why Perform MRI After EEG Spikes in a Child?

When a child’s EEG shows epileptic spikes localized to a particular cortical region (for example, left cortical spikes seen during sleep), physicians often recommend an MRI scan. The rationale is that focal EEG abnormalities might be caused by an underlying structural brain lesion such as focal cortical dysplasia (FCD) or another malformation. MRI is the imaging modality of choice to investigate such cases, as it can reveal subtle cortical malformations that might be triggering the seizures. In other words, the EEG tells us something is wrong electrically in a region, and MRI is used to look for something wrong structurally in that same region.

Importantly, performing an MRI after detecting focal EEG spikes helps rule out serious causes like tumors or large malformations, and it can confirm the presence of FCD or other abnormalities if present. For instance, international guidelines advise obtaining an MRI for anyone diagnosed with focal epilepsy, especially in children, unless the epilepsy is a known idiopathic syndrome (like benign centrotemporal spikes) where structural imaging is usually normal. In the case of benign childhood epilepsy with centrotemporal spikes (BECTS), the EEG shows characteristic spikes during sleep in the centrotemporal (rolandic) region, but the brain development is normal – thus MRI in such cases is expected to be normal and often not required unless the diagnosis is uncertain. In summary, an MRI after EEG spikes is done to find any hidden structural cause of the epilepsy and guide appropriate treatment (for example, planning epilepsy surgery if a lesion like FCD is found).

What Can a “Normal” MRI Tell Us (and Its Limitations)?

Sometimes, the MRI comes back normal (showing no obvious lesion) even though the EEG had spikes. A normal MRI is certainly reassuring in that it rules out large or obvious brain lesions, but it does not entirely exclude the presence of FCD or other subtle cortical malformations. In fact, not all FCDs are visible on MRI – many histologically confirmed FCDs can be so subtle that conventional MRI appears normal. This is especially true for milder forms like FCD Type I, which often produce no clear MRI changes. Thus, an MRI report of “normal” or “no significant abnormality” means no clear lesion was detected by standard imaging, but small FCDs could still be present (these cases are sometimes called “MRI-negative” FCD).

It’s important to understand the limitations of MRI resolution and human detection: FCD lesions can be very small or blending in with normal brain tissue. They often hide in complex folds of the cortex and may only cause very subtle signal changes on a few MRI slices. A busy radiologist might understandably miss such faint abnormalities. In fact, surgical studies have shown that 30–50% of patients who were thought to have “MRI-negative” epilepsy actually have a lesion like FCD upon surgical histopathology examination. In other words, many children with normal MRI scans do turn out to have FCD when more advanced imaging or surgery is done – the MRI just wasn’t able to spot it. The normal MRI is still useful: it establishes a baseline and helps ensure no large lesion was overlooked, and it prompts clinicians to perhaps use other tools (like high-resolution MRI, specialized post-processing, PET scans, or repeat imaging) if clinical suspicion for FCD remains high.

👉 Key point: A normal MRI does not guarantee the absence of FCD; it might simply mean the dysplasia is too subtle for routine MRI to detect. This is why having the MRI reviewed by expert neuroradiologists is recommended in tough cases, and why researchers are developing advanced imaging techniques to uncover what looks normal at first glance.

MRI Features of Focal Cortical Dysplasia

On MRI scans, FCDs (when visible) often produce characteristic structural abnormalities in the brain. Understanding these helps explain what radiologists look for:

• Cortical thickening: The affected area of cortex may be thicker than normal due to abnormal layering of neurons.
• Blurred gray-white matter junction: Instead of a sharp boundary between cortex (gray matter) and underlying white matter, FCD can blur this interface. This happens because mislocated neurons extend into the white matter, disrupting the normal architecture. The MRI appears “hazy” at that spot.
• Abnormal MRI signal: The dysplastic region often shows hyperintensity on T2/FLAIR sequences. This can appear in the cortex itself and/or in the subcortical white matter. In some FCDs, a hallmark “transmantle sign” is seen – a tapering column of T2/FLAIR signal extending from the cortical lesion down toward the ventricle. This reflects a radial orientation of the abnormal tissue through the brain’s thickness.
• Unusual gyral/sulcal pattern: The folding of the brain may look irregular or disrupted in the dysplastic area. FCD can cause a local malformation of cortical development, sometimes manifesting as a shallow sulcus or an oddly shaped gyrus.
• Focal volume changes: Some FCDs lead to slight atrophy or hypoplasia of the affected segment of brain, meaning that part of the brain may be slightly shrunken or underdeveloped. This is the result of long-standing structural disorganization.
• No mass effect or enhancement: Unlike a tumor, FCDs do not typically cause swelling, edema, or take up contrast dye. They are developmental anomalies, not growing masses, so they usually blend in without pushing on other structures.

These MRI features are the “science behind” how FCD presents: they reflect the underlying developmental disorganization of brain tissue. For example, the blurring of the gray-white junction corresponds to abnormal neurons in the wrong place (white matter), and the transmantle sign corresponds to a column of dysplasia spanning from the cortex inward. Radiologists carefully inspect MRI scans for these telltale signs, often using high-resolution MRI protocols dedicated to epilepsy, to catch even subtle dysplasias.

Types of FCD and Their MRI Characteristics

Neurologists and pathologists classify focal cortical dysplasias into subtypes (per the International League Against Epilepsy classification by Blumcke et al.) – and each subtype tends to have different visibility on MRI:

• FCD Type I (mild cortical dysplasia): This involves abnormal cortical lamination (layers of neurons). MRI appearance: Type I is often very subtle or invisible on MRI. Many Type I cases show no clear lesion – at most, there may be a slight blurring of the gray-white boundary or a subtle signal change, often in the temporal lobe. Because Type I is a microscopic dysplasia of the cortical architecture, standard MRI frequently looks normal.
• FCD Type IIa: This subtype has dysmorphic neurons in the cortex (more severe disorganization). MRI appearance: Type IIa lesions usually can be seen on MRI. They often present as a region of cortical thickening with an indistinct gray-white junction. The cortex may look enlarged or swollen and doesn’t cleanly separate from the white matter. One might also notice an abnormal gyral pattern in that area.
• FCD Type IIb: This is a more severe form (sometimes called “Taylor dysplasia”) with dysmorphic neurons andballoon cells. MRI appearance: Type IIb shows all the features of IIa (blurred junction, cortical thickening, etc.) plus the classic transmantle sign – a streak of abnormal signal extending from the cortical lesion towards the ventricle. In fact, a transmantle sign is seen in about 94% of Type IIb cases. This makes Type IIb the easiest to identify on MRI; the combination of thick cortex, fuzzy boundaries, and a radial band into white matter is highly suggestive of FCD. Radiologists often specifically look for this sign when reviewing MRI of epilepsy patients.
• FCD Type III: This category is used when an FCD is found alongside another brain lesion (for example, Type IIIa is FCD with hippocampal sclerosis, IIIb is FCD adjacent to a tumor like a ganglioglioma, IIIc with a vascular malformation, IIId with an old injury/glial scar). MRI appearance: In FCD Type III, the MRI is usually dominated by the other lesion. The dysplasia itself may not be obvious because changes from the primary pathology (scarring, tumor, etc.) overshadow it. For instance, if a dysplasia is next to a tumor, the MRI will clearly show the tumor, and the FCD might only be recognized in retrospect.

Knowing these types is useful because it explains why some FCDs show up readily on scans (Type IIb being the most conspicuous) whereas others are “MRI-negative” (Type I especially). When an MRI is read as normal but the child’s EEG and symptoms strongly suggest a focal epilepsy, a Type I FCD or very subtle Type II could be the culprit lurking beneath the radar of standard imaging.

EEG Spikes vs. MRI Findings: Putting the Puzzle Together

EEG and MRI provide complementary information. EEG spikes indicate the electrical irritability of a region, while MRIshows the structural anatomy. In many cases of FCD-related epilepsy, the EEG spikes will localize to the same area where MRI finds a lesion, reinforcing the diagnosis. For example, if sleep EEG shows repeated spikes over the left frontal region and MRI reveals a focal cortical dysplasia in the left frontal cortex, we have a match that likely explains the child’s seizures.

However, when EEG and MRI don’t line up (e.g. EEG points to left cortex but MRI looks normal), it becomes a diagnostic challenge. A normal MRI despite focal spikes could mean: (a) the lesion is too subtle to see, (b) the epilepsy might be one of the syndromes that truly have no structural lesion (idiopathic epilepsy), or (c) the EEG focus might be mis-localized due to technical factors. In practice, if the clinical picture doesn’t fit a benign syndrome, doctors assume there may indeed be a small lesion and sometimes pursue more sensitive investigations. For instance, they might do MRI post-processing analyses (computer-aided techniques to highlight cortical abnormalities), or use other imaging like PET or SPECT to find metabolic clues of a hidden lesion. In some specialized centers, high-field 7 Tesla MRI is used for children with otherwise MRI-negative focal epilepsy. Research has shown that 7T MRI (especially with advanced image post-processing) can uncover FCD lesions that were missed on standard 3T scans. In one report, adding 7T MRI plus a morphometric analysis program detected subtle cortical lesions in over one-third of patients who previously had “normal” 3T MRIs. Many of these newly found lesions were confirmed as FCD and corresponded exactly to the area of EEG abnormality. This illustrates that when EEG clearly points to a focal area (for example, continuous spikes in the left cortex) and routine MRI is normal, more refined techniques or expert reviews often reveal the hidden dysplasia.

In short, EEG spikes guide us where to look, and MRI tells us what’s there. If MRI shows nothing obvious, that information is still useful: it tells us we’re dealing with either an MRI-negative lesion or a non-structural epilepsy. The care team can then decide on further steps (observation, medical treatment, or deeper diagnostics) based on the whole picture.

How AI and Deep Learning Are Enhancing Diagnosis and Therapy

The advent of artificial intelligence (AI) and deep learning is proving to be a game-changer for conditions like FCD-related epilepsy – benefiting researchers, clinicians, and families alike. These technologies are being applied to both EEG and MRI analysis, as well as treatment planning, to improve accuracy and outcomes:

• AI in MRI lesion detection: One of the hardest problems has been detecting very subtle FCDs on MRI. AI algorithms (including machine learning models and deep neural networks) can be trained on brain scans to spot the minute textural or shape anomalies that a human might overlook. In fact, several AI-based tools for FCD detection have been developed (e.g. MAP, DeepFCD, MELD classifiers), aiming to flag potential dysplastic lesions automatically. These tools are especially valuable in cases where everything looks normal but the clinical evidence (like focal EEG spikes or PET scan abnormalities) strongly suggests an FCD is present. By letting an algorithm pore over the MRI with superhuman thoroughness, subtle changes in cortical thickness or signal can be identified. This means more children with “MRI-negative” epilepsy might finally get a diagnosis of FCD if AI highlights a suspicious area for doctors to investigate. Early studies show that AI models can achieve high sensitivity in detecting known FCD lesions, though improving their specificity (reducing false alarms) remains an ongoing challenge. As these models are refined, they hold promise for assisting neuroradiologists – essentially acting as a second pair of eyes that never get tired. In the near future, an epileptologist could feed a patient’s MRI into an AI system and be pointed to a tiny cortical dimple or blur that warrants closer review, potentially shortening the time to diagnosis.
• AI in EEG analysis: Reading EEGs is a labor-intensive task for neurologists, often likened to finding needles (spikes or seizures) in hours of squiggly haystack. Deep learning is now being harnessed to automate EEG interpretation. Sophisticated neural network models (CNNs, RNNs, etc.) can learn to recognize the patterns of epileptic discharges in raw EEG data. This offers several benefits: it can detect epileptic spikes or seizures quickly and accurately, and it doesn’t get fatigued by long recordings. Studies have demonstrated that deep learning models can sift through EEG data and highlight segments with seizures or interictal spikes with high sensitivity, often matching or exceeding human expert performance. What this means for clinicians is faster diagnosis and the ability to monitor patients in real-time or over long durations (like 24/7 at home) with an automated alert system. For researchers, having AI analyze EEGs provides a consistent way to quantify brain activity and perhaps discover new biomarkers (because the AI might pick up subtle waveform features that humans consider just “noise”). Ultimately, this could lead to earlier diagnosis – for example, an algorithm might recognize a dangerous seizure pattern in an infant’s EEG that a generalist might miss, prompting life-saving treatment sooner.
• Seizure prediction and personalized therapy: Beyond identifying seizures, AI is being explored for predictingwhen a seizure might occur before it happens. By analyzing EEG trends or other data, machine learning models attempt to forecast impending seizures, which could enable warning systems or preemptive therapies. This is still an evolving area, but early results are promising. Additionally, AI can help in personalizing treatment plans. Epilepsy treatment often involves trial-and-error with medications; AI algorithms are being developed to analyze a patient’s specific data (seizure patterns, genetics, imaging, etc.) and suggest which antiseizure medication or therapy might work best for that individual. In drug-resistant cases, AI might assist in identifying which patients would benefit from epilepsy surgery by integrating imaging, EEG, and neuropsychological data to predict surgical outcomes. All these applications aim to take some of the guesswork out of therapy and make care more tailored to the person.
• AI-powered patient monitoring and support: Another area where AI is helping both clinics and families (especially mothers and caregivers) is in seizure monitoring devices. For example, researchers are testing smart monitoring systems that use AI to detect seizures from video and audio in a child’s bedroom. One such system (the Nelli® system) uses a camera and microphone by the bedside; its AI algorithms analyze the child’s movements and sounds during sleep to determine if a seizure is occurring. This kind of technology can immediately alert parents or caretakers when a nocturnal seizure is detected, bringing peace of mind and enhancing safety. The child doesn’t need to wear any sensors – the AI “watches” over them silently. Such tools are incredibly valuable for parents who worry about unattended seizures at night. They also generate reports for physicians, so doctors can review how many seizures happened and adjust medications accordingly. In broader terms, AI-driven wearable devices (like smartwatches or EEG headbands) and home monitoring systems are bridging the gap between hospital and home care, ensuring that seizures are caught and counted accurately. This eases the burden on families and provides clinicians with reliable data to guide treatment.

In summary, AI and deep learning are enriching every step of epilepsy care: from detecting the hidden lesion on MRI, to analyzing EEGs faster and more consistently, to predicting seizures and optimizing treatment, and even to keeping watch over patients in their daily lives. For researchers, these technologies open new frontiers to understand epilepsy by crunching massive datasets and uncovering patterns (for example, finding subtle imaging biomarkers of FCD that were never noted before). For clinicians, AI acts as a powerful assistant – increasing diagnostic yield (finding what was once missed) and saving time – which ultimately leads to more confident decision-making. And for mothers and families, innovations like AI seizure monitors and more precise diagnostics translate to peace of mind and hopefully better outcomes. As one review aptly stated, AI offers promising solutions by improving the accuracy and efficiency of epilepsy diagnosis and management. It’s an exciting era where cutting-edge computer science is being harnessed to improve the lives of children with conditions like focal cortical dysplasia, ensuring they get the best possible diagnosis, treatment, and care.

Sources

1. Radiopaedia – Focal Cortical Dysplasia (MRI features and types)
2. NICE Guideline NG217 – Epilepsies in children (imaging recommendations)
3. Cleveland Clinic (Consult QD) – 7T MRI for MRI-negative epilepsy (FCD detection at 7T)
4. Walger et al., Epilepsia (2023) – AI in FCD detection (role of AI when EEG suggests FCD but MRI is negative)
5. Mallick et al., Frontiers in Neuroscience (2024) – Deep learning for EEG seizure detection (automated EEG analysis advantages)
6. AbuAlrob et al., Epilepsy & Behavior (2025) – Review on AI in epilepsy (AI for EEG analysis, seizure prediction, personalized treatment)
7. Le Bonheur Children’s Hospital – AI for Epilepsy (Nelli\u00ae system) (AI seizure monitoring to help parents)
8. exas Children’s Hospital – Epileptic Encephalopathies (details on LKS and CSWS/ESES symptoms, diagnosis, and treatment)texaschildrens.orgtexaschildrens.orgtexaschildrens.org.
9. Radiopaedia – Focal Cortical Dysplasia (MRI features and types of FCD)radiopaedia.orgradiopaedia.orgradiopaedia.org.
10. Radiopaedia – Tuberous Sclerosis (Cortical Tubers) (association of tubers with epilepsy and autism traits)radiopaedia.orgradiopaedia.org.
11. Epilepsy Foundation – Polymicrogyria (effects on development and diagnosis by MRI)epilepsy.comepilepsy.com.
12. Medline/StatPearls – Lennox-Gastaut Syndrome (characteristic EEG and developmental impact)rarediseases.orgiranepilepsy.org.
13. Research literature on MRI-negative FCD detection and multimodal epilepsy workupradiopaedia.orgsciencedirect.com.