This session covered a number of topics in stroke imaging that ranged from carotid/aortic imaging, young stroke recurrence, CTP thresholds, perviousness and computerized interpretation using artificial intelligence/machine learning algorithms.
Several presentations highlighted potential novel imaging risk markers.
Carotid plaque inflammation imaged with FDG-PET (Camps-Renom) showed an association with early stroke recurrence. However, most patients with recurrence in the study had carotid stenosis >50%, leaving uncertainty as to how the carotid inflammation finding could help with risk stratification and treatment.
Proximal aortic arch stiffness on 4D flow MRI (Wehrum) was associated with large (>4mm) aortic plaques. Aortic stiffness was also more common in stroke patients compared to controls. Small cortical DWI lesions were found in 1.8% of young stroke cases at MRI follow-up many years after stroke, and this was also associated with markers of cerebral small vessel disease, suggesting alternative potential mechanisms for cerebral SVD (van Dongen).
High signal lesions on MR “black blood” imaging of carotid arteries had significant less short-term residual stenosis despite no post-stent balloon dilatation during the initial procedure (Tanno). This could imply absorption of plaque haemorrhage with time, lessening the degree of stenosis.
Several studies described the performance of software based on artificial intelligence paradigms, concluding prognostic value from automated ASPECTS in a selected group of ENCHANTED cases (Nagel), good agreement with expert detection of large artery occlusion (Barreira) or collateral scoring (Pfaff). Some future role for radiologists might remain amidst all of the automation, at least as beta testers for software.
Andrew Bivard presented a case-control study that sought to critically evaluate the current CBF threshold for ischaemic core prediction in the context of much more rapid and reliable reperfusion with endovascular therapy, concluding that a threshold of CBF <20% of contralateral optimally predicted irreversible injury (with AUC 0.89 compared to 24h DWI) when reperfused rapidly with endovascular treatment, compared to the 30% threshold derived from intravenous thrombolysis studies.
Manon Kappelhof used pooled HERMES data to investigate thrombus perviousness, concluding that higher perviousness of the occlusive thrombus (measured as thrombus attenuation increase, TAI, on co-registered CTA compared to non-contrast CT) was linked to favourable outcome with intravenous thrombolysis , but not with endovascular treatment. The parameter may identify non-responders to IV therapy. Further evaluation combining perviousness measurements with other indices such as clot length will clarify its future role.