Combination of CT perfusion parameters to improve definition of brain ischemic core

Acute ischemic stroke is a leading cause of disability and mortality, so it is of extreme importance to accurately define the infarct core (irreversibly damaged tissue) from the penumbra (potentially salvageable tissue), after the stroke, to guide treatment decisions. CT Perfusion (CTP) imaging provides key hemodynamic parameters for infarct core estimation: Cerebral Blood Volume, Cerebral Blood Flow, Mean Transit Time, and Time to Maximum. However, traditional single-parameter threshold methods often fail to effectively differentiate the ischemic core from surrounding penumbra. A considerable part of this project consists in developing a pipeline that transforms CT perfusion images in numerical data, allowing to perform a quantitative study that explores novel combinations of perfusion parameters to enhance ischemic core definition. A dataset of 18 patients was analyzed, utilizing statistical comparisons, Principal Component Analysis (PCA), and unsupervised clustering techniques. Several parameter combinations were tested to improve core delineation. Results confirm that single-parameter thresholds effectively distinguish infarcted from healthy tissue but struggle to separate the core from the penumbra. Among the proposed combinations, those integrating all parameters with a simple formula showed potential in refining core definition. Additionally, correlation and clustering analyses incorporating clinical data did not yield significant trends, but raised the possibility of distinguishing patients with different ischemic status. In the last part of the project, an example of perfusion map of the combination of parameters is created and, after applying a threshold to it, compared to the commonly used single-value threshold approach, showing an improvement in performance. This research highlights the limitations of traditional thresholding methods and suggests that multi-parameter models could improve ischemic core assessment.