Prerequisite of SPECT/CT image segmentation approach potentially leading to precise absolute quantification of myocardial focal tracer uptake: Phantom and canine validations

Abstract

Absolute quantification from cardiac single photon emission computerized tomography (SPECT) remains challenging due to the complication of heterogeneous background and extracardiac radioactivity, causing a substantial estimation error in the quantification of radiotracer uptake in the heart. The partial volume effect resulted from the limited SPECT resolution is another confounding source contributing to this estimation error. Precise segmentation of the myocardial edges is an imperative prerequisite for reliable corrections of the extracardiac activity and partial volume errors. This paper is to introduce a unique CT segmentation method developed based on the level-set framework and to demonstrate the feasibility of this method for the detection of the myocardial edges. The image segmentation method developed was validated using a cardiac phantom and in vivo canine model. The endo- and epi-cardial edges from CT images of the cardiac phantom and the dog's heart were segmented successfully using the level-set-based image segmentation approach. We demonstrated in this study that the segmentation algorithm was feasible to precisely delineate the endo- and epi-cardial edges, providing the important anatomical information about the heart for the corrections of partial volume and extracardiac activity errors and subsequently improving the absolute SPECT quantification of molecularly targeted tracer uptake in the myocardium.