Abstract
Purpose: CT angiography-source image (CTA-SI) can be used as an effective alternative to diffusion-weighted imaging (DWI) for identifying acute ischemic stroke (AIS). This study investigates the reliability of combining CTA-SI with non-contrast CT (NCCT) for AIS diagnosis, with a focus on how different brain areas affect diagnostic accuracy. Methods: Patients with various subtypes of AIS who underwent NCCT, CTA, and DWI from January to December 2022 were included. Two experienced neuroradiologists analyzed ischemic core across NCCT, CTA-SI, and NCCT+CTA-SI models, evaluating interobserver reliability and lesion detection rate. Results: A total of 304 patients (63% male, age 67.2 ± 11.9 years) with AIS were included. The distribution of stroke subtypes was as follows: 23% large vessel trunk infarction, 46% deep perforator vessel infarction, 9% superficial perforator vessel infarction, 5% watershed infarction, and 17% infratentorial infarction. The interobserver reliability was substantial in the three image models, especially the NCCT+CTA-SI model (all p<0.05). The NCCT+CTA-SI model demonstrated higher lesion detection rate than the NCCT (59.20% vs 48.7%, p<0.05) and CTA-SI model (59.2% vs 45.4%, p<0.05), particularly when detecting large vessel trunk infarction (82.90% vs 58.60%, p<0.05) and deep perforator vessel infarctions (64.80% vs 44.40%, p<0.05). Conclusions: The NCCT+CTA-SI model may be a valuable tool for evaluating AIS when DWI is not feasible. Smaller hospitals might consider adopting this combination for improved stroke diagnosis, highlighting the need for careful evaluation of deep perforator vessel infarction when large vessel trunk infarction is not evident.