238. Accuracy and precision of a novel algorithm for T1 and extracellular volume fraction mapping in cardiac magnetic resonance imaging. (December 2018)

• Record Type: Journal Article
• Authors: Borrazzo, C.; Galea, N.; Carnì, M.; Francone, M.; Catalano, C.; Carbone, I.; Pacilio, M.
• Is Part Of: Physica medica. Volume 56(2018)Supplement 2
• Page Start: 209
• Page End: 210
• Publication Date: 2018-12
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2018.04.249
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : Purpose: Extracellular Volume Fraction (ECV) can diagnose cardiac disorders, and is obtainable using T1 mapping. The sensitivity for ECV quantification is limited by the uncertainty affecting T1 measurements. A new methodology for T1 and ECV mapping was here tested, estimating also its level of uncertainty. Methods: A reference phantom containing test objects with known T1 values was used, and 60 subjects (40 diseased and 20 healthy) were examined pre- and post-administration of the contrast agent, using a 1.5 T MR tomograph and MOLLI sequences, . A novel algorithm based on a Fast Non-linear Least Squares (FNLS)[1] method was implemented for modelling MR signals, and compared with other well-established methodologies. Gamma-tool analysis, with tolerances criteria of 3% for T1 difference and 4.21 mm for distance-to-agreement was also used to evaluate differences between measured and reference T1 maps. Results: Bland-Altman plots and gamma tool analysis evidenced the good agreement among methods. Differences of mean T1 measured by test objects and FNLS algorithm (with respect to reference values) were in the range −3%/2%. The average and maximal standard deviation were 2.3% and 7.4% for the reproducibility with time, and 1.5% and 2.4%for the T1 spatial reproducibility. The ECV error (in terms of error propagation) ranged between 1.0–3.5%. The inter-patient variability range of the ECV value was 23.4%–29.8% for the healthy controls, and 31.0%–53.9% for the diseased patients, showing significant differences between the two groups (p = 0.012). Conclusion: The FNLS algorithm is suitable to achieve accurate T1 and ECV mapping. The values of ECV referred to healthy controls were in very good agreement with previously published results. ECV uncertainty, in terms of propagation of T1 errors, were limited and comparable with the healthy control inter-patient variability, confirming the accurate ECV quantification.