Abstract Synthetic MRI offers the advantage of reducing acquisition time and enhancing flexibility through the reconstruction of various contrast weightings from a single set of MRI scans. However, the use of synthetic T1-weighted (synT1-w) MRI can lead to potentially biased measurements of the cross-sectional area (CSA) in the spinal cord when compared to conventionally acquired T1-weighted MRI. This disparity can have implications for comparability and sensitivity of MRI in assessing disease progression or treatment effects in neurodegenerative spinal cord disorders. Thus, this study aimed at improving the accuracy (i.e., difference between synthetic and acquired MRI) of cervical cord CSA measurements (C1-C3 level) based on synT1-w MRI implementing a longitudinal data set acquired from 23 acute spinal cord injury (SCI) patients and 21 healthy controls over 2 years. Moreover, the validity of using synT1-w MRI for tracking cervical cord atrophy following SCI over 2 years was verified. SynT1-w images were reconstructed from quantitative maps of proton density, longitudinal, and effective transverse relaxation rates derived from a multi-parameter mapping protocol. The results showed a minimal bias of -0.31 mm2 (-0.5%) in CSA measurements based on synT1-w compared to acquired MRI. Estimates of atrophy rates and average CSA were comparable between synthetic and acquired MRI. A sample size estimation for detecting treatment effects on CSA atrophy after 2 years following SCI revealed that the required sample size is reduced by 13.5% using synT1-w instead of acquired MRI. This study shows high accuracy of synT1-w MRI and demonstrates its applicability in clinical studies for optimizing long MRI protocols.
Read full abstract