PT  - JOURNAL ARTICLE
AU  - Cai, Yuxin
AU  - Zhang, Jianhai
AU  - Arvind, Ganesh
AU  - Hu, Bo
AU  - Bijoy, Menon
AU  - Chen, Shengcai
AU  - Qiu, Wu
TI  - CT Perfusion Map Generation from Multi-phase CT Angiography Using Generative adversarial model for Acute Ischemic Stroke
AID  - 10.3174/ajnr.A8857
DP  - 2025 May 29
TA  - American Journal of Neuroradiology
PG  - ajnr.A8857
4099  - http://www.ajnr.org/content/early/2025/05/29/ajnr.A8857.short
4100  - http://www.ajnr.org/content/early/2025/05/29/ajnr.A8857.full
AB  - BACKGROUND AND PURPOSE: Multiphase CT Angiography (mCTA) has shown potential as a diagnostic tool for acute ischemic stroke (AIS), as it captures dynamic changes in cerebral vasculature. However, mCTA has limitations in assessing brain tissue perfusion, which reduces its clinical interpretability. To address this limitation, we aim to develop a generative adversarial network (GAN) that generates CT Perfusion (CTP)-like maps from mCTA. This approach aims to improve the interpretability of mCTA.MATERIALS AND METHODS: A total of 714 cases with NCCT, CTP, mCTA, and follow-up NCCT/MR were analyzed across internal and external datasets. A GAN was trained to generate multi-parametric CTP maps (Tmax, CBF, CBV). The model's performance was evaluated using SSIM, PSNR, and FID compared to actual CTP maps. Clinical utility was assessed by predicting infarct core and penumbra using threshold-based segmentation and evaluating metrics such as Dice coefficient, AUC of dichotomized infarct volume of < 70cc and mismatch ratio following DEFUSE 3 criteria, compared to the ground truth of actual CTP prediction.RESULTS: The GAN achieved SSIM 0.647–0.662, PSNR 20.6–20.9, and FID 16.6–17.0 on internal data, surpassing both CycleGAN [11] (SSIM: 0.608–0.642, PSNR: 18.2–19.2, FID: 27.6–32.5) and Pix2Pix [10] (SSIM: 0.630–0.645, PSNR: 19.5–19.7, FID: 19.4–20.8) across all metrics. Predicted penumbra and infarct core showed Dice coefficients of 0.672 and 0.468, with strong correlations (penumbra: 0.921, core: 0.902) and AUCs of 0.854 (95% CI: 0.819–0.888)(mismatch ratio) and 0.850(95% CI: 0.817–0.884) (dichotomized infarct core). External data validation yielded Dice coefficients of 0.481 (penumbra) and 0.301 (core) with AUCs of 0.720(95% CI: 0.589– 0.808) (mismatch ratio) and 0.703(95% CI: 0.528–0.794)(dichotomized infarct core).CONCLUSIONS: The GAN effectively generated CTP-like maps from mCTA, improving interpretability and demonstrating promising diagnostic performance, particularly for resource-limited settings.ABBREVIATIONS: mCTA = Multiphase CT angiography, CTP = CT perfusion, CBF = Cerebral blood flow, CBV = Cerebral blood volume, GAN = Generative adversarial network, FID = Fréchet inception distance, AUC = Area under the receiver operating characteristic curve, AIS = acute ischemic stroke