Myocardial strain measurements using magnetic resonance feature tracking proves to be a new promising tool for cardiac function assessment, but only a few studies address intervendor consistency, especially for segmental strain.
Our aim was to evaluate segmental strain measurement reproducibility and accuracy among 4 different software.
We selected 45 patients, appointed to 3 groups: 15 normal, 15 with dilated cardiomyopathy and 15 with acute infarction and performed measurements of longitudinal, circumferential and radial strain employing software from Medviso, Circle, TomTec and Medis. Reproducibility between strain values obtained with the 4 different vendors was assessed by comparing software among each other as well as with the mean of all of them. Furthermore intra- and inter-observer variability was assessed with interclass correlation coefficient (ICC) and coefficient of variation (CV). Finally, the accuracy of strain measurement for detection of structural (infarcted segments) as well as functional pathology (septal vs. lateral wall strain in left bundle branch block (LBBB)) was assessed.
There was significant intervendor difference in segmental strain for longitudinal, circumferential and radial strain (all three p < 0.001). Medviso demonstrated the best intra- as well as inter-observer variability with ICC > 0.962 and CV <24%, followed by Medis and TomTec with ICC > 0.812 and CV < 54%. Circle displayed acceptable ICC (0.715) but high coefficient of variability (109%). There was significant difference between vendors also in the discrimination of infarcted segments, or septum dysfunction in LBBB patients. In this case Circle displayed the highest difference btw pathologic and normal segments, with best performance for longitudinal strain measurements.
Intervendor differences for segmental strain measurements are significant rendering some software approaches more reproducible than others and stressing the need for further standardization.