Background: axial aortic strain has prognostic value for aneurysm risk stratification1, which is particularly important in patients with Marfan syndrome. Advances in three-dimensional dynamic aortic imaging enable displacement-based strain estimation2,3, but such approaches often neglect strain directionality, overlooking important biomechanical information3,4. We developed a fully automatic method to quantify local biaxial strains in aortic geometries derived from balanced steady-state free-precession magnetic resonance imaging. Method validity was evaluated by testing the literature-supported hypothesis that axial strain exceeds circumferential strain5. Method: a pre-trained deep learning segmentation model2 was used to automatically segment thoracic aortas in 9 Marfan syndrome patients and 9 healthy volunteers across 30 cardiac phases. End-diastolic configuration (X) was registered to the remaining phases (x_phase) to obtain displacement fields (U_phase=x_phase-X) (Figure). Subsequently, the deformation gradient tensor (F_phase^elem=(∂x_phase)⁄∂X) was derived per element. Axial direction (A^elem | |A^elem |=1) in the reference configuration was defined by assigning geometry centerline tangent vectors to each element centroid. Circumferential direction (C^elem | |C^elem |=1) was then computed as (A^elem x N^elem), where N^elem is the surface normal. For each phase and element, the axial (ε_(ax,phase)^elem) and circumferential strains (ε_(circ,phase)^elem) were calculated (Figure) by (|〖F_phase^elem∙A〗^elem |-1) and (|〖F_phase^elem∙C〗^elem |-1). Per patient, max┬phase⁡〖(ε_(ax,phase)^elem)〗⁡and max┬phase⁡〖(ε_(circ,phase)^elem)〗⁡were collected. Paired and unpaired t-tests were performed on the collected data to study the difference between the maximum value of ε_ax or ε_circ within group and between groups. Results and conclusion: in both groups, axial strain was higher than circumferential strain (patients: ε_ax= 0.36 ± 0.14 vs ε_circ = 0.33 ± 0.11, p = 0.098; healthy: ε_ax= 0.35 ± 0.11 vs ε_circ = 0.34 ± 0.11, p = 0.083), corroborating previous findings5. No between-group differences were observed (pax = 0.964, pcirc = 0.897). Future work will include higher-powered studies and regional strain averaging, in addition to maximum values, to better distinguish healthy and diseased aortas.