Influence of Region-Of-Interest Definition and Image Source on Test-Retest Repeatability of Myocardial T1 and T2...

Native T1- and T2-mapping provide quantitative myocardial tissue biomarkers. For diffuse myocardial disease, mid-ventricular septal region of interest (ROI) is recommended, but evidence comparing this approach with full-slice strategies, contour erosion, and source-image fitting remains limited.

Twenty healthy volunteers underwent two separate 1.5 T magnetic resonance imaging (MRI) examinations on the same day. Native T1 MOLLI 5(3)3 and T2-prepared balanced steady state free precession (bSSFP) mapping were acquired in three short-axis (SAx) slices (basal, mid, apical). Two cardiovascular radiologists performed two independent reading sessions, using both source images and inline maps. Septal and full-slice measurements were obtained, with increasing contour erosion depths applied. Intra-reader, inter-reader, test-retest, and cross-reader test-retest agreement was assessed using intraclass correlation coefficient (ICC)(2,1). Mixed-effects models estimated baseline values, session and reader biases, and smallest detectable differences (SDDs).

Mean baseline T1 was 1001 ms (CoV 1.5%) and T2 48 ms (CoV 4.4%). Septal ROI showed poorest reproducibility: test-retest ICCs 0.625-0.654 (T1) and 0.438-0.653 (T2); cross-reader ICCs 0.401-0.553 (T1) and 0.188-0.221 (T2); SDDs reached up to 57.5 ms (5.9%) for T1 and 3.84 ms (9.0%) for T2. Optimal T1 precision occurred with full-slice source-image analysis excluding apical slice (SDD 36.6 ms, 3.7%; ICC 0.837). For T2, full-slice inline maps achieved highest test-retest (0.903-0.915) and cross-reader ICCs (0.901-0.910); increasing erosion from 0% to 40% reduced SDD from 1.9 ms to 1.6 ms.

Post-processing choices materially impact myocardial T1 and T2 repeatability in healthy volunteers. Septal ROI sampling is less reproducible than full-slice strategies for longitudinal and cross-reader use.