1.1 Research materials
This is a retrospective study. The clinical data and MRI of 273 thalassemia patients in the First Affiliated Hospital of Guangxi Medical University from January 2011 to December 2015 were collected. The inclusion criteria were: (1) Patients were genetically diagnosed thalassemia and had a regular history of blood transfusion. (2) 9 years old ≤ age ≤ 50 years old. (3)Patients had both the T2* sequence MRI with intact liver 12 echoes and the Ferriscan LIC report in the corresponding period (the required T2 image scan time for FerriScan is the same as the corresponding T2* image scan time, and the difference between the FerriScan LIC reporting time and the corresponding T2* measurement time does not exceed 48 hours). The exclusion criteria were: (1) MRI artifacts were too large to meet the measurement requirements. (2) Patients had other chronic liver diseases or tumor diseases. Among the 273 patients included, there were 152 males and 121 females, aged from 9 to 49 years old, with an average of (21.1 ± 10.53) years old. All patients (or parents/guardians) gave written informed consent to participate in the study.
1.2 Mr Scanning Methods
MRI was performed on a 1.5T scanner (MAGNETOM Avanto Fit, Siemens Healthcare, Erlangen, Germany).
T2* data were acquired using a multiecho GRE scanning sequence at the same level above the liver gate at the last breath (Hold breath once and scan for the same level above the hilum). Relevant pulse sequence parameters include: flip angle = 20°, echo time(TE) = 1.29, 2.35, 3.43, 4.6, 5.68, 6.85, 7.93, 9.1, 10.18, 11.35, 12.43, 13.6 ms, repetition time (TR) = 200.00 ms, FOV read = 400 mm, matrix = 256, layer thickness = 10 mm. Scan time was 15s.
The FerriScan acquisition consisted of a free-breathing 2D multislice spin-echo pulse sequence. Relevant pulse sequence parameters include:flip angle = 90°, echo time (TE) = 6, 9, 12, 15, 18 ms, repetition time (TR) = 1000 ms, FOV read = 400 mm, matrix = 256, and 11 slices of 5 mm thickness.
1.3 Data Measurement And Analysis
T2 image data was sent to FerriScan for processing and analysis. The LICF used in this study was derived from the final FerriScan report. As mentioned before, the required T2 image scan time for FerriScan is the same as the corresponding T2* image scan time, and the difference between the FerriScan LIC reporting time and the corresponding T2* measurement time does not exceed 48 hours.
The T2* image data were all post-processed by CMRtools(CMRtools/Thalassemia Tools, Cardiovascular Imaging Solutions, London, UK). Measurement process: the anonymous image was imported into the software and, avoiding the intrahepatic blood vessels and bile ducts seen by naked eyes at the same level of the liver, the roughly same ROI were drawn according to the area measured by FerriScan. And the drawn ROI and matching T2* values appeared in the post-processing software. Truncation method [14, 15] was used to discard the interference signal values deviating from the fitting curve, and T2* value was recorded when the determination coefficient value (R2) ≥ 0.98 (Fig. 1).
1.4 Statistical Methods
Statistical analysis was performed using SPSS 26.0 statistical software package. Using a random number table, 273 cases of T2* data were divided into 191 cases in the test group and 82 cases in the verification group according to the ratio of 7:3. The measured liver T2* values did not conform to the normal distribution by the normality test. By curve fitting (exponential function), the T2* values of the test group and LICF were used to build the calibration curve equation.
The T2* value in the verification group was converted to LICe by the equation. After inspection, LICe and LICF did not conform to the normal distribution. Friefmans M test was used to explore the difference between them. If P > 0.05, there was no significant difference. Kendall's coefficient was used to explore the consistency between them. If the correlation coefficient W > 0.75 and P < 0.05, it means there is a high degree of consistency. Spearman rank correlation analysis was used to explore the correlation between the two. If the correlation coefficient |rs|>0.75 and P < 0.05, it means there is a high correlation.
According to the 1.5T MRI grading of liver iron concentration, LICe and LICF were divided into normal group (0.17-1.8mg/g dry weight), mild group (1.8-7.0mg/g dry weight), moderate group (7.0-14.0mg/g dry weight), severe group (> 14.0 mg/g dry weight). The McNemar test was used to explore the difference of the clinical grading results of LICe and LICF. If P > 0.05, there was no significant difference between them. The Kappa's coefficient was used to explore the consistency of clinical grading results of them. If K > 0.75 and P < 0.05, it means that there is high consistency between them.