[Accuracy of different segmentation methods to quantify cardiac function parameters and left ventricular mass in an...

To evaluate accuracy of manual, semiautomatic and automatic segmentation methods to quantify cardiac function parameters, volumes, and left ventricular mass.

Ten Goettingen miniature pigs (seven male, three female; mean weight 49.8 kg; SD: 10.65; range: 35-65 kg) were evaluated using a 1.5 Tesla MR scanner and free-breathing SSFP sequences. Short axis images were employed to quantify cardiac parameters. In all animals volumes (end-diastolic volume, end-sisstolic volume, stroke volume), ejection fraction and left ventricular mass was calculated. To quantify these parameters three different segmentation methods (manual, semiautomatic, and automatic) were employed. The correlation of different segmentation methods for each variable was analyzed.

No statistically significant differences in left ventricular volume or mass quantification using different segmentation methods was found (p > 0.05). Correlation to estimate ejection fraction and end-diastolic volume was good (r > 0.6) and turned to be excellent for the rest of the variables (r > 0.8). Mean quantification time was less for semiautomatic and automatic methods than for the manual method.

SSFP sequences allow to accurately evaluate cardiac function parameters and left ventricular mass in a large animal model with any of the segmentation method described in this report. Non-manual segmentation methods allow rapid quantification of cardiac parameters.