Background Myocardial deformation can be key to clinical decision. 2D feature-tracking of cardiovascular magnetic resonance (CMR-FT) imaging is user-friendly but has poor reproducibility, particularly for radial strain. 3D CMR-FT may improve repeatability by reducing through-plane artefact. The aim of this study was to provide normal ranges for 3D CMR-FT and compare its reproducibility to older generation 2D CMR-FT. Method 56 asymptomatic, healthy subjects (43.7±12.9 year, 52% male) undertook CMR (1.5 Tesla scanner Magnetom Avanto, Siemens, Erlangen, Germany). 3D FT-CMT was generated using the SSFP HLA, VLA and short axis cine images (Figure 1). A single observer (BL) analysed the CMR studies using 2D and 3D CMR-FT (Circle cvi42® version 5.3) and after 4 weeks, re-analysed blinded scans for intra-observer variability. Inter-observer variability was generated by separate tracking by a second blinded observer (AS) in a randomly generated subset of 15 subjects. Agreement was tested by calculating mean bias and 95% limits of agreement (confidence intervals) from Bland–Altman analyses, coefficient of variation, and inter-class correlation coefficient (ICC). Results There is modest agreement between all measures of 2D and 3D peak strain analysis (ICC=0.44 to 0.58). Mean global circumferential strain (GCS) on 3D analysis is −16.8±2.5, compared to 2D GCS of −20.2±3.31 and −20.6±3.4 at the base and mid-ventricular level respectively. Mean global longitudinal strain (GLS) is −13.7±2.3 on 3D CMT-FT and −19.3±2.7 on 2D CMR-FT. Global radial strain (GRS) is 45.5±10.9 for 3D, compared to 55.1±14.4 and 48.0±13.4 at the base and mid-ventricular level respectively. Table 1 displays the inter- and intra-observer variability of each technique. Intra-observer variability was significantly improved by 3D CMT-FT for GCS, whilst inter-observer variability was significantly improved for GCS, GRS and strain rates. No reproducibility differences were identified for GLS. Discussion Peak strains using 3D FT-CMR is different to 2D normal range values. 3D CMR-FT has superior intra- and inter-observer reproducibility compared with 2D CMR-FT, particularly for GCS and GRS strain, the latter being the principal systolic strain and should improve detection of sub-clinical ventricular dysfunction.Abstract 111 Table 1 Intra- and inter-observer reproducibility for 2D and 3D CMR-FT Variability Mean 3D bias±SD Mean 2D bias±SD T-Test on 2D vs 3D bias 3D intraclass correlation coefficient (95% CI) 2D intraclass correlation coefficient (95% CI) GLS Intra-observer 1.73±1.43 1.85±1.52 p=0.70 0.64 (0.38 to 0.79) 0.67 (0.50 to 0.79) GLS Inter-observer 1.11±1.48 1.58±1.91 p= 0.25 0.53 (0.04 to 0.82) 0.62 (0.19 to 0.85) GCS Intra-observer 1.10±1.10 1.87±1.74 p< 0.01 0.81 (0.67 to 0.89) 0.75 (0.61 to 0.85) GCS Inter-observer 1.02±1.22 2.08±2.28 p= 0.03 0.79 (0.79 to 0.88) 0.56 (0.04 to 0.83) GRS Intra-observer 6.66±8.06 8.90±9.71 p= 0.10 0.63 (0.44 to 0.77) 0.61 (0.42 to 0.75) GRS Inter-observer 6.13±7.48 10.78±12.67 p= 0.04 0.67 (0.27 to 0.87) 0.35 (-0.09 to 0.71)Abstract 111 Figure 1 Generation of 3D CMR-FT model through HLA, VLA and short axis contours