AUTHOR=Luo Chaotian , Peng Fei , Xu Fengming , Tang Cheng , Zhang Yanyan , Huang Chaojie , Liang Linlin , Ning Xiaojing , Peng Peng 

TITLE=Assessing the accuracy of CMRtools software for diagnosing liver iron overload in thalassemia patients: influencing factors and optimisation strategies

JOURNAL=Frontiers in Medicine

VOLUME=Volume 11 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2024.1424294

DOI=10.3389/fmed.2024.1424294

ISSN=2296-858X

ABSTRACT=Background: CMRtools is a software package that can be used to measure T2* values to diagnose liver iron overload, however, its accuracy in terms is affected by multiple factors, including goodness-of-fit (R 2 value), the number of echo time (TE) images, and the liver iron concentration (LIC). To investigate the effects of the R 2 value, the number of TE images, and the LIC on the accuracy of CMRtools software for measuring T2* values to diagnose liver iron overload (LIO).CMRtools software was used to measure liver T2* values among 108 thalassemia patients via the truncation method, and the R 2 values, the number of TE images, and T2* values were recorded. These values were subsequently converted into liver iron concentration (LICT) values. The LICF (derived from MRI-R2/FerriScan) was used as a reference, and the diagnostic accordance rate (DAR) was compared between R 2 value subgroups, between TE image number subgroups, and between LIC subgroups.The greater the R 2 value was, the greater the standardized DAR (SDAR) was (P < 0.05). The SDAR are not identical between each TE image number subgroup (P > 0.05). However, the relationship between TE image number subgroups and SDAR was analysed using Spearman's correlation, and it was found to be positively correlated (rs = 0.729, P = 0.017). The SDAR are not identical between each LIC subgroup (P > 0.05), furthermore, the relationship between LIC subgroup and SDAR was found irrelevant (P = 0.747).The accuracy of CMRtools software for diagnosing LIO in patients with thalassemia can be improved by artificially controlling the number of TE images to be fitted and selecting higher R 2 values.