Quantification of cardiac iron in patients with thalassemia with 3-T MRI calibrated by 1.5-T MRI.

BACKGROUND: Due to the small sample size of many studies, it remained unclear what standardized reference range the T2* cutoff at 3 T would be used to assess the severity of cardiac iron load. In addition, the number of patients with moderate to severe cardiac iron load was small in some studies, especially the sample of patients with severe cardiac iron load. PURPOSE: To explore the feasibility, reproducibility, and reliability of using T2* values in quantifying cardiac iron load in patients with thalassemia at 3 T. MATERIAL AND METHODS: A total of 122 patients with thalassemia underwent cardiac T2* imaging at both 1.5 T and 3 T. Cardiac R2* (1000/T2*) values of the 100 patients at 3 T were fitted against the values at 1.5 T using linear regression and the prediction equation was derived. The remaining 22 cases were used to test the prediction accuracy of the equation. RESULTS: The combined R2* values exhibited a strong linear relationship between 1.5 T and 3 T (r = 0.830，P<0.001). At the center, it had a slope of 1.348 and an intercept of 37.279. According to the equation, the truncated T2* values of cardiac iron overload and cardiac heavy iron overload at 3 T were <10 ms and <6 ms, respectively. The two truncated T2* values were used to diagnose different levels of cardiac iron overloaded of 22 patients at 3 T; the accuracy rates were 95.5% and 100.0%, respectively. CONCLUSION: T2* quantification of cardiac iron load at 3 T MRI resulted to be feasible, reproducible, and reliable.

α- and ß-Genotyping of Thalassemia Patients Based on a Multimodal Liver MRI Radiomics Model: A Preliminary Study in Two Centers.

BACKGROUND: So far, there is no non-invasive method that can popularize the genetic testing of thalassemia (TM) patients on a large scale. The purpose of the study was to investigate the value of predicting the α- and ß- genotypes of TM patients based on a liver MRI radiomics model. METHODS: Radiomics features of liver MRI image data and clinical data of 175 TM patients were extracted using Analysis Kinetics (AK) software. The radiomics model with optimal predictive performance was combined with the clinical model to construct a joint model. The predictive performance of the model was evaluated in terms of AUC, accuracy, sensitivity, and specificity. RESULTS: The T2 model showed the best predictive performance: the AUC, accuracy, sensitivity, and specificity of the validation group were 0.88, 0.865, 0.875, and 0.833, respectively. The joint model constructed from T2 image features and clinical features showed higher predictive performance: the AUC, accuracy, sensitivity, and specificity of the validation group were 0.91, 0.846, 0.9, and 0.667, respectively. CONCLUSION: The liver MRI radiomics model is feasible and reliable for predicting α- and ß-genotypes in TM patients.