A new clinical tool for the quantification of myocardial CT perfusion imaging in patients with suspected Ischemic Heart Disease

Abstract

The standard assessment of myocardial perfusion has several limitations since it is mainly based on its relative distribution. In the clinical practice, the evaluation of perfusion by using Computed Tomography (CT) Imaging is usually performed visually or semi-quantitatively. The scarcity of quantitative perfusion data not always allows a proper diagnose of patients which are suspected of suffering from some diseases, such as Ischemic Heart Disease (IHD). This project aims to develop a clinical tool for the automatic quantification of myocardial perfusion in patients with suspected IHD. Myocardial perfusion is assessed based on a combined diagnosis protocol (CT/CTP protocol) which involves the acquisition of two contrast-enhanced CT images, one obtained at rest and another acquired under pharmacological stress. All perfusion images are evaluated in the short-axis view of the left ventricle (basal, medium and apical slices), divided in different segments according to the 16-AHA-segmentation model and divided circumferentially in two layers with the same thickness to define subendocardium and subepicardium. The clinical tool, which is developed with MATLAB, enables the automatic quantification of perfusion in each myocardial segment by providing the mean of Hounsfield Units in those regions. Based on this analysis the clinicians can compared the values at baseline and at hyperemia, and they can better determine hypoperfusion defects in patients with IHD. The validation of the clinical tool is performed by comparing automatic and manual perfusion measurements of 10 patients with suspected IHD who were previously assessed with Single Photon Emission Computed Tomography (SPECT) for perfusion analysis. A strong linear correlation is found between the automatic and manual results. Afterwards, perfusion defects obtained from CT/CTP protocol are compared to perfusion defects from SPECT, to assess the applicability of this clinical tool for the diagnosis of IHD. Sensitivity, specificity, positive predictive value and negative predictive values of CT/CTP protocol are 0.64, 0.36, 0.48 and 0.88, respectively. Results from these measurements are not quite good. Nonetheless, the analysis should be done increasing the number of patients to have a better assessment of this protocol for the diagnosis of this disease.