Cortical spreading depolarizations (CSDs) are large-scale disruptions of neuronal homeostasis that contribute to secondary injury after ischaemic stroke and occur across a range of acute and chronic neurological disorders. In healthy cortex, CSDs are typically coupled to compensatory vasodilation, whereas in ischaemic tissue they can evoke inverse haemodynamic responses that exacerbate metabolic stress and promote lesion progression, highlighting the need for approaches that can resolve localised electrophysiological and vascular dynamics as a function of evolving tissue metabolic state.

Here, we combine high-density graphene micro-transistor arrays with optical cerebral blood flow and oxygenation imaging to map perfusion-dependent electrophysiological and haemodynamic dynamics of CSDs during focal cerebral ischaemia. Using two complementary stroke models, we show that CSD waveform duration and morphology scale systematically with local perfusion gradients, revealing distinct electrophysiological signatures that reflect tissue metabolic state. These signatures resolve a continuum from relatively preserved cortex to markedly ischaemic regions within a single preparation. We identify a characteristic CSD waveform that precisely localises profoundly hypoperfused tissue and demonstrate that specific electrophysiological features predict whether a given CSD will elicit vasodilatory, biphasic, or vasoconstrictive haemodynamic responses. Using neural network modelling, we further show that DC-coupled electrophysiology alone can accurately predict local vascular responses to CSDs, indicating that critical information about tissue state and neurovascular risk is encoded within the CSD waveform itself in these models. Concurrent oxygenation imaging reveals a focal pre-onset decline in tissue oxygenation at sites of CSD initiation and shows that prolonged CSDs impose sustained metabolic burden, linking electrophysiological dynamics to tissue oxygen recovery. Finally, we demonstrate that low-dose ketamine shortens CSD duration, thereby preventing the inverse haemodynamic response in metabolically compromised tissue, providing a physiological mechanism for its reported neuroprotective effects. Together, these findings establish perfusion-dependent CSD signatures as high-fidelity functional biomarkers of tissue state and neurovascular risk, enabling real-time delineation of severely compromised, at-risk, and relatively preserved cortex. While demonstrated here in ischaemic stroke, this framework has broad relevance for monitoring injury evolution and therapeutic response in neurological disorders where spreading depolarizations play a pathogenic role.