Validation of fully automated myocardial T1 and synthetic hematocrit-based extracellular volume quantification on 3T cardiac MRI: a retrospective cross-sectional study
Year: 2026
Type: article
Abstract: Background: Automated post-processing software has been developed to quantify various cardiac magnetic resonance (CMR) parameters, including native T1 and extracellular volume (ECV) using synthetic hematocrit (Hct). However, validation studies of these automated measurements remain limited. We aimed to validate a fully automated workflow integrating myocardial T1 analysis and synthetic Hct for ECV quantification on 3T CMR, using manual ECV calculation based on venous Hct as the reference method. Methods: This retrospective, cross-sectional study included 65 adults with low-to-intermediate risk of atherosclerotic cardiovascular disease who underwent stress 3T CMR. Manual ECV was calculated using manually contoured myocardial T1 values and venous Hct obtained within 7 days of CMR. Automated analysis used Myomics-T1 software (Phantomics, Korea) to automatically generate myocardial and blood-pool T1 values and contours, with synthetic Hct derived from a regression model developed in an independent cohort of 450 participants. Correlation and agreement were assessed using Pearson correlation, intraclass correlation coefficients (ICC), and Bland-Altman analysis. Results: 27.8%±4.6%). Synthetic Hct correlated with venous Hct (r=0.78; ICC =0.76) but demonstrated proportional bias. Agreement between synthetic Hct-derived and manual ECV was better within the normal Hct range (35-45%) but substantially reduced at Hct >45% (ICC =0.04). Conclusions: Fully automated ECV quantification demonstrated strong correlation and agreement with the manual method within the normal Hct range. However, synthetic Hct-derived ECV was less reliable at abnormal Hct levels, particularly at Hct >45%. Further external validation and refinement of the synthetic Hct model are warranted to support broader clinical application.
Language: English
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7221747570