We evaluated a mathematical algorithm for the generation of medullary signal from raw dynamic magnetic resonance (MR) data. Five healthy volunteers were studied. MR examination consisted of a run of 100 T1-weighted coronal scans (gradient echo; TR/TE 11/3.4 msec, flip angle 60 degrees; slice thickness 6 mm; temporal resolution 2 seconds). Gadolinium-diethylene triamine pentaacetic acid (DTPA; 0. 05 mmol/kg) was injected with an injector pump (5 ml/sec). Medullary MR renographs (MRRs) were calculated for regions of interest with strong and moderate cortical volume averaging (CVA). A reference medullary MRR, devoid of CVA, was obtained. Percentual signal differences between calculated and reference medullary MRRs were estimated for each consecutive scan. Run averaged values of these differences were calculated. Mean values, after subtraction of the resting state signal, were +0.2% (SD 9.7%) and +0.7% (SD 9.0%) for areas with strong and moderate CVA, respectively. We conclude that with this algorithm reliable extraction of medullary MRRs is feasible, providing a unique tool for clinical evaluation of medullary disease. J. Magn. Reson. Imaging 2000;12:453-459.
Copyright 2000 Wiley-Liss, Inc.