Is magnetic resonance imaging the 'reference standard' for cardiac functional assessment? Factors influencing measurement of left ventricular mass and volumes

Purpose: MRI is considered reference standard for the assessment of left ventricular (LV) volume and mass measurements. There are few accepted guidelines for uniform assessment of cardiac function with MRI. We sought to investigate different confounding factors influencing LV measurement results. Material and Methods: In 60 diabetic type-II patients (group A) we compared intra-/inter-reader variability of MRI for cardiac function measured twice at a 3 month interval by one MRI trained reader and one untrained. In 20 patients (group B) two different techniques were compared for determining the epicardial and endocardial LV-borders. Results: Bland Altman analysis showed excellent intra-observer measurement agreement for the trained reader 1 for EDM (mean=-2.3 (-23.6-19)), EDV (2.9(-9.2-15.0)), ESV (3.3(-5.8-12.4)) and EF (1.2(-3.3-5.7)). Untrained reader 2 measurement agreement was considerably less appropriate for EDM (mean=-8.2 (-25.8-9.5)), EDV (7.8(-5.1-20.7)), ESV (5.3(-8.0-18.6)). Only for EF (0.8 (-6.5-8.1)) results were comparable to reader 1. Inter-observer measurement in the beginning was poor for EDM (-13.5(-55.6-28.6)) and EDV (7.3(-61.9-76.6)), whereas agreement for ESV (2.1(-29.9-34.2)) and EF (-0.9(-11.6-9.9)) was good. After 3 months, measurement agreement for EDM (-5.3 (-46.4-35.8)) was considerably improved, for EDV (0.4(-67.0-66.2)) was excellent, whereas agreement for ESV (3.1(-34.4-28.1)) and EF (-1.7(-13.0-9.6)) was similar. Using different techniques for determining the epicardial and endocardial borders, only end-diastolic volume was unchanged whereas all other parameters were significantly different using the two methods (p ≤ 0.03). Conclusion: Intra- and inter-reader variability, analyst experience as well as different techniques for determining the boundaries of the left ventricle significantly affect MRI parameters for cardiac function. These results suggest a need for developing commonly accepted standards for cardiac MRI evaluation.