The impact of semi-automatic versus manually adjusted assessment of global longitudinal strain in post-myocardial infarction patients

There are unresolved questions related to the proper use of editing the region of interest (ROI) for measurements of global longitudinal strain (GLS). The purpose of the present study was to compare the semi-automatic default GLS value by the vendor’s software with manually adjusted GLS and test the impact on GLS measures with different ROI widths. We selected 25 patients post myocardial infarction treated with PCI who had excellent echocardiographic recordings after 2–5 days and 3 months. The different GLS values were assessed from these 50 analyses in three steps. The semi-automatically GLS by default ROIs was compared with manually adjusted ROIs widths selected by an expert and then with manual adjustments, but with fixed ROIs being narrow, medium and wide. Their mean age was 64 (± 12) years, 52% had ST elevation MI and mean LVEF was 52 (± 4)%. Mean default GLS was − 15.3 (± 2.5)% with the widest ROI level selected semi-automatically in 78% of all widths. The mean expert GLS with manually adjusted ROI was − 14.7 (± 2.4)%, and the medium ROI level was selected by the expert in 85% of all examinations. The mean adjusted GLS, but with fixed ROIs widths was − 15.0 (± 2.5%)% with narrow ROI, − 14.7 (± 2.6)% with medium and − 13.5 (± 2.3)% with wide ROI width (p < 0.001 vs. default GLS). The Intra Class Coefficient Correlation between default and manually adjusted expert GLS was 0.93 (p < 0.001). The difference between the default and the manually adjusted expert GLS was neglectable. These findings may represent a simplification of the assessment of GLS that might increase its use in clinical practice. The GLS measurements with a fixed wide ROIs were significantly different from the expert measurements and indicate that a wide ROI should be avoided. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1007/s10554-020-01826-4) contains supplementary material, which is available to authorized users.