Integrated EEG Medicine

From waveforms to a clinical instrument

Everything in the preceding pillars converges here. The expert does not see an EEG as a set of squiggles to be matched against a memorized catalogue; the expert sees a real-time, spatially resolved readout of cortical network dynamics and interrogates it the way a physiologist interrogates any complex system — by forming hypotheses about generators, states, and pathology, then testing them against the record across montages, time, and clinical context. This final module formalizes that act into a reproducible sequence so that nothing is omitted under time pressure and so that the reasoning remains explicit, auditable, and teachable.

The eleven-step interpretive sequence

Apply the following sequence to every study. Earlier steps constrain later ones — you cannot judge whether a sharp transient is epileptiform (step 5) before you have characterized the background it interrupts (steps 1-2), and you cannot estimate prognosis (step 9) without first establishing reactivity and continuity (steps 1-2) in their clinical setting (step 6).

1. Identify the background activity. Establish the posterior dominant rhythm, its frequency, and the anterior-posterior gradient. The background is the canvas against which every abnormality is judged.
2. Characterize frequency and organization. Quantify dominant frequencies, symmetry, continuity, variability, and reactivity to stimulation. Disorganization and loss of reactivity are early, sensitive markers of dysfunction.
3. Identify abnormalities, if present. Separate the record into background versus superimposed transients, rhythms, and discontinuities, and describe each by morphology, field, and temporal behavior.
4. Localize cortical generators. Use phase reversal in bipolar chains and amplitude maxima in referential montages to assign a generator to each abnormality; cross-check across montages.
5. Distinguish epileptiform from non-epileptiform activity. Demand the full morphologic and field criteria before assigning epileptiform status; actively consider benign variants and artifacts as competing hypotheses.
6. Correlate with the clinical state. Reconcile the electrographic state with the documented behavioral state, medications, and the clinical question. Discordance is itself a finding.
7. Determine seizure risk. Translate interictal findings and history into a calibrated statement of recurrence or first-seizure risk, acknowledging the limited sensitivity of a single routine study.
8. Assess encephalopathy severity. Grade diffuse dysfunction by the degree of slowing, disorganization, and the presence of patterns such as triphasic waves, and track its trajectory.
9. Estimate prognosis, where applicable. In the ICU and post-arrest setting, integrate continuity, reactivity, and malignant patterns into a prognostic formulation, explicitly multimodal and never in isolation.
10. Recommend the next diagnostic or therapeutic step. State what the record implies for management: repeat or prolonged recording, sleep-deprived study, continuous monitoring, imaging, or a treatment trial.
11. Quantify uncertainty. Attach a calibrated confidence to each conclusion and name what would change it. A report that hides its uncertainty is less safe than one that states it.

Bayesian discipline at every node

The sequence is not a checklist to be completed mechanically; each step is a place where a prior meets evidence. The base rate of epilepsy in the population referred for a routine outpatient EEG is low, so a borderline sharp transient should move your posterior far less than the same transient in a patient with stereotyped focal seizures. The corollary is the central error of the field: a confident clinical history of spells inflates the prior and seduces the reader into over-calling sharp-looking normal variants. Discipline means keeping the morphologic evidence and the clinical prior separable in your mind, so that you can state how much each contributed to the conclusion — and so that a colleague reading the same record could reconstruct your reasoning.

A worked synthesis

Return to the real electrographic seizure below and run the full sequence aloud: establish the interictal background and its organization, mark the moment new rhythmic activity breaks that background, localize the onset by its field, confirm it is ictal by its evolution in frequency and spatial extent rather than by amplitude alone, reconcile it with the clinical state, and only then translate it into seizure risk and the next step. The same eleven steps that organize a normal outpatient study organize the most urgent ICU record; only the priors and the stakes change.

The final competency

At completion, EEG is no longer interpreted as isolated waveforms. It becomes a real-time readout of cortical network dynamics, a physiological measure of brain state, a diagnostic instrument for epileptogenesis, a prognostic tool in coma and critical illness, and a computational signal-processing system grounded in neurobiology. The waveform is merely the surface; what you are reading is the brain.