We present an anthropomorphically accurate left ventricular (LV) phantom derived from human computed tomography (CT) data to serve as the ground truth for the optimization and the spatial resolution quantification of a CT-derived regional strain metric (SQUEEZ) for the detection of regional wall motion abnormalities. Displacements were applied to the mesh points of a clinically derived end-diastolic LV mesh to create analytical end-systolic poses with physiologically accurate endocardial strains. Normal function and regional dysfunction of four sizes [1, 2/3, 1/2, and 1/3 American Heart Association (AHA) segments as core diameter], each exhibiting hypokinesia (70% reduction in strain) and subtle hypokinesia (40% reduction in strain), were simulated. Regional shortening (RS_CT) estimates were obtained by registering the end-diastolic mesh to each simulated end-systolic mesh condition using a nonrigid registration algorithm. Ground-truth models of normal function and of hypokinesia were used to identify the optimal parameters in the registration algorithm and to measure the accuracy of detecting regional dysfunction of varying sizes and severities. For normal LV function, RS_CT values in all 16 AHA segments were accurate to within ±5  %  . For cases with regional dysfunction, the errors in RS_CT around the dysfunctional region increased with decreasing size of dysfunctional tissue.