The clinical value of magnetic resonance perfusion imaging (MRI) was investigated by quantitative comparison with (201)thallium-single-photon emission computed tomography ((201)TI-SPECT) and quantitative coronary angiography (QCA). Short-axis imaging was performed during dipyridamole administration in 13 patients with single-vessel coronary artery disease. Using inner and outer contours, the myocardium was divided into 30 contiguous, radial regions. Defining a perfusion defect as a region with less than 90% of maximum (201)TI intensity, nine patients had a matching perfusion defect, two had no defect on both (201)TI-SPECT or MRI, and one had a defect on (201)TI-SPECT but not on MRI. One patient had a defect on both modalities but with inaccurate localization. Three perfusion parameters were investigated: a) maximum contrast enhancement (MCE); b) slope of the signal intensity versus time curve; and c) inverse mean transit time (1/MTT). The sensitivity and specificity of MCE in the detection of perfusion abnormalities with TI-SPECT as the reference method were 71% and 71%, respectively (slope 77% and 61%, 1/MTT 44% and 70%). Furthermore, correlations were calculated per patient for the entire circumference of the short-axis myocardium. Median correlations were as follows: MCE 0.92, slope 0.91, and 1/MTT 0.40. Mismatches between (201)TI defects and defects on MRI resulted in low mean correlations (MCE 0.45, slope 0.46, and 1/MTT 0.26). There was a trend between severity of perfusion defects on MRI (using MCE) and QCA stenosis area (r = -0.56, P = 0.06). Thus, MRI and (201)TI-SPECT demonstrate fair agreement in the assessment of perfusion defects but show moderate correlation when the entire short-axis myocardium is correlated.
Copyright 2000 Wiley-Liss, Inc.