Improved parametric mapping of CT perfusion in acute ischemic stroke

Computed Tomography Perfusion (CTP) imaging is crucial for assessing cerebral blood flow in acute ischemic stroke, but traditional deconvolution methods like Singular Value Decomposition (SVD) often lack physiological accuracy and are sensitive to noise. This study introduces a clustering-based deconvolution framework integrated with the Adiabatic Approximation to Tissue Homogeneity (AATH) model to address these limitations. By grouping voxels with similar timedensity curves and applying the AATH model, our method incorporates physiological constraints and reduces noise, leading to more accurate estimates of perfusion parameters such as cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), and time to maximum (Tmax). Simulations showed that the approach outperforms the SVD method, particularly in recovering accurate parameter maps even under mismatched model assumptions. The method's robustness and potential for improved clinical outcomes, such as better differentiation between salvageable and irreversibly damaged tissue, highlight its value for precision diagnostics in cerebrovascular diseases.