Comparing manual and AI assisted methods for stroke in non-contrast CT imaging

Comparing manual and AI assisted methods for stroke in non-contrast CT imaging

Product: e-ASPECTS Company: Brainomix


Comparative Analysis of the ABC/2 Score and e-ASPECTS Software in the Determination of Acute Ischaemic Stroke Volume from Non-Contrast CT

brain sciences, 2025

Abstract

Background and Purpose

Accurate and reproducible methods for assessing infarct volume in acute ischaemic stroke have important therapeutic and prognostic implications for the choice and success of acute therapeutic interventions. However, there is no international consensus on the methodology employed in infarct volume assessment. We aimed to assess the reliability of the ABC/2 score and e-ASPECTS in the determination of infarct volume in acute ischaemic stroke.

Methods

Infarct volume was measured from NCCT in stroke patients recruited ≤12 h of symptoms onset and at 24 h using the ABC/2 method. Automated ischaemic volume measurements were carried out using e-ASPECTS software. Measurements using ABC/2 were compared with e-ASPECTS to assess volume differences and reliability using Lin’s concordance correlation coefficient.

Results

Thirty-three patients with CT < 12 h from onset of symptoms and follow-up at 24 h were included in the analysis. Use of ABC/2 demonstrated low agreement between observers (0.490, CI 0.236–0.743, p < 0.001) on admission (<12 h). High agreement was found between observers at 24 h (0.724, CI 0.564–0.884, p < 0.001). High agreement was observed between the mean observed infarct volumes using ABC/2 and e-ASPECTS on admission (0.794, CI 0.691–0.898, p < 0.001).

Conclusions

Our results suggest that e-ASPECTS is a reliable platform for ischaemic volume determination particularly in the hyperacute phase to inform management. However, the use of ABC/2 represents an alternative approach to e-ASPECTS in the rapid and reliable estimation of ischaemic infarct volume to inform prognosis and treatment decisions, particularly in cases of delayed presentation where infarction is established and arterial territory boundaries are easily identifiable.

Read full study