Quantitative Magnetic Resonance First-Pass Perfusion Analysis: Inter- and Intraobserver Agreement

Abstract

Magnetic resonance first-pass (MRFP) imaging awaits longitudinal clinical trials for quantification of myocardial perfusion. The purpose of this study was to assess inter- and intraobserver agreement of this method. Seventeen MRFP studies (14 rest and 3 under adenosine-induced hyperemia) from 14 patients were acquired. Two observers visually graded study quality. Each study was subdivided into eight regions. Both observers analyzed all 17 studies (8 x 17 = 136 regions) for interobserver agreement. Each observer then analyzed 10 of the 17 studies a second time (2 x 8 x 10 = 160 regions) for intraobserver agreement. Signal intensity curves were obtained with Argus software (Siemens, Iselin, NJ). The maximum amplitude of the impulse response function (Rmax) and the change of signal intensity (deltaSImax) of the contrast bolus were determined. Intraclass correlation coefficient was used to determine intra- and interobserver agreement. The quality was good or excellent in 14 studies. Intraobserver agreement of Rmax and deltaSImax were good (0.85 and 0.80, n = 160). Interobserver agreement of Rmax was fair (0.55, n = 136) but improved after exclusion of poor-quality studies (0.88, n = 112). Interobserver agreement of deltaSImax was good (0.73) and improved less than Rmax with study quality (0.83). Interobserver agreement for Rmax in individual myocardial regions before and after exclusion of studies with poor quality changed most markedly in lateral and posterior regions (0.69 and 0.65 vs. 0.97 and 0.94), where signal-to-noise ratios were reduced compared with anteroseptal regions (p < 0.01). Analysis of MRFP images provides good intraobserver agreement. Interobserver agreement of the quantitative perfusion analysis is good under the premise of good image quality.

Direct comparison of myocardial perfusion cardiovascular magnetic resonance sequences with parallel acquisition

“…However, many studies have not necessarily had a suitable clinical dimension being limited by using doped phantoms, not comparing sequences in the same patients, performing only rest perfusion studies, or not systematically assessing the presence and severity of artifacts (33,34). In our study, hEPI scoring the least artifacts and the highest diagnostic confidence is consistent with the findings of Muhling et al (35), who demonstrated that interobserver agreement as well as sensitivity to perfusion defects is related to both image quality and SNR. In this study, a technical sequence comparison, we could not confirm that hEPI was associated with significantly improved diagnostic clinical performance because the very small sample size (15 patients) was insufficient for showing diagnostic differences.…”

Evidence for Microvascular Dysfunction in Hypertrophic Cardiomyopathy

“…The use of MRI in the present study provided the important advantage of spatially matched, high-resolution measurements of LV hypertrophy, MBF, and myocardial fibrosis all made within 1 MRI study of Ͻ60 minutes in duration. [25][26][27] We found reduced hMBF in HCM patients with and without hypertrophy. This is consistent with PET data demonstrating impaired microvascular function in both hypertrophied and nonhypertrophied myocardium in such patients, 10 although the resolution of PET is suboptimal for assessing the magnitude of hypertrophy.…”

“…The majority of our flow data in both healthy volunteers and HCM patients at rest and during hyperemia are in agreement with previous studies using first-pass MRI and other modalities, such as PET. 5,10,27 Slight deviations may be explained by differences in baseline characteristics, such as age, gender distribution, and blood pressure, because all of these substantially influence MBF. 34 In agreement with data from Camici and colleagues, 10 we confirmed that the MBF at rest in HCM was not different from that in healthy controls, and this finding was independent of wall thickness, myocardial fibrosis, and other potential confounders.…”

Multislice first‐pass myocardial perfusion imaging: Comparison of saturation recovery (SR)‐TrueFISP‐two‐dimensional (2D) and SR‐TurboFLASH‐2D pulse sequences

“…Mü hling et al (22) demonstrated that interobserver agreement as well as sensitivity to perfusion defects is related to image quality and SNR. This corresponds to findings in five patients of our study in whom a perfusion defect could be observed with TrueFISP but not with Turbo-FLASH.…”