Quantitative pixel-wise measurement of myocardial blood flow: The impact of surface coil-related field inhomogeneity and a comparison of methods for its correction

BACKGROUND Surface coil-related field inhomogeneity potentially confounds pixel-wise quantitative analysis of perfusion CMR images. This study assessed the effect of surface coil-related field inhomogeneity on the spatial variation of pixel-wise myocardial blood flow (MBF), and assessed its impact on the ability of MBF quantification to differentiate ischaemic from remote coronary territories. Two surface coil intensity correction (SCIC) techniques were evaluated: 1) a proton density-based technique (PD-SCIC) and; 2) a saturation recovery steady-state free precession-based technique (SSFP-SCIC). METHODS 26 subjects (18 with significant CAD and 8 healthy volunteers) underwent stress perfusion CMR using a motion-corrected, saturation recovery SSFP dual-sequence protocol. A proton density (PD)-weighted image was acquired at the beginning of the sequence. Surface coil-related field inhomogeneity was approximated using a third-order surface fit to the PD image or a pre-contrast saturation prepared SSFP image. The estimated intensity bias field was subsequently applied to the image series. Pixel-wise MBF was measured from mid-ventricular stress images using the two SCIC approaches and compared to measurements made without SCIC. RESULTS MBF heterogeneity in healthy volunteers was higher using SSFP-SCIC (24.8 ± 4.1%) compared to PD-SCIC (20.8 ± 3.0%; p = 0.009), however heterogeneity was significantly lower using either SCIC technique compared to analysis performed without SCIC (36.2 ± 6.3%). In CAD patients, the difference in MBF between remote and ischaemic territories was minimal when analysis was performed without SCIC (0.06 ± 0.91 mL/min/kg), and was substantially lower than with either PD-SCIC (0.50 ± 0.63 mL/min/kg; p = 0.013) or with SSFP-SCIC (0.63 ± 0.89 mL/min/kg; p = 0.005). In 6 patients, MBF quantified without SCIC was artifactually higher in the stenosed coronary territory compared to the remote territory. PD-SCIC and SSFP-SCIC had similar differences in MBF between remote and ischaemic territories (p = 0.145). CONCLUSIONS This study demonstrates that surface coil-related field inhomogeneity can confound pixel-wise MBF quantification. Whilst a PD-based SCIC led to a more homogenous correction than a saturation recovery SSFP-based technique, this did not result in an appreciable difference in the differentiation of ischaemic from remote coronary territories and thus either method could be applied.