Artificial Neural Network Computer Tomography Perfusion Prediction of Ischemic Core

Area of Science:

• Neuroimaging
• Artificial Intelligence in Medicine
• Stroke Imaging

Background:

• Computed tomography perfusion (CTP) is crucial for assessing acute ischemic stroke.
• CTP aids in estimating the ischemic core and penumbra.
• Optimal CTP parameters for ischemic core identification are not yet established.

Purpose of the Study:

• To determine optimal artificial neural network (ANN) parameters for predicting ischemic core.
• To evaluate ANNs using diffusion-weighted imaging (DWI) as the gold standard.
• To compare ANNs trained on CTP data versus combined clinical and CTP data.

Main Methods:

• Developed ANNs to predict ischemic core volume in acute stroke patients.
• Utilized diffusion-weighted imaging (DWI) as the reference standard.
• Trained two ANNs: one with CTP data alone, another with clinical and CTP data.

Main Results:

• CTP-based ANN achieved a mean absolute error (MAE) of 13.8 mL (SD 13.6 mL) vs. DWI.
• Combined clinical and CTP data yielded similar MAE (13.8 mL) but lower SD (12.4 mL).
• The combined ANN showed improved performance metrics (AUC 0.87, sensitivity 0.91, specificity 0.65) compared to CTP-only (AUC 0.85, sensitivity 0.90, specificity 0.62).

Conclusions:

• ANNs effectively predict the ischemic core in acute stroke.
• Integrating clinical data with CTP data improves the accuracy and reliability of ischemic core prediction.
• This approach offers a more precise method for evaluating stroke severity and guiding treatment.