Scientific publications

Dual-source CT in heart transplant recipients: quantification of global left ventricular function and mass

May 1, 2009 | Magazine: Journal of Thoracic Imaging

Bastarrika, Gorka MD, PhD (1); Arraiza, Maria MD (1); De Cecco, Carlo Nicola MD (3); Broncano, Jordi MD (1); Mastrobuoni, Stefano MD, PhD (2); Ubilla, Matias MD (2); Pueyo, Jesus C. MD, PhD (1); Rabago, Gregorio MD (2)
Institution Departments of (1) Radiology
(2) Cardiovascular Surgery, Clinica Universitaria, Universidad de Navarra
(3) Department of Radiology, Universita' di Roma "Sapienza," Ospedale Sant' Andrea.

To evaluate reliability of global left ventricular (LV) function and mass quantification in heart transplant recipients undergoing cardiac dual-source computed tomography examinations by means of manual contour tracing and using a region-growing-based semiautomatic segmentation analysis software tool.

Twenty-six consecutive heart transplant recipients undergoing cardiac dual-source computed tomography examinations with tube current modulation were included. Double-oblique short-axis 8-mm slice thickness multiphase image reconstructions were used for manual contouring and axial 0.75 mm slices were used for the semiautomated segmentation. LV ejection fraction, end-diastolic volume, end-systolic volume, stroke volume, cardiac output, and myocardial mass were assessed by both segmentation methods. Length of time needed for manual contour tracing and for the semiautomated software was recorded. Contrast-to-noise ratio was calculated in end-diastolic and end-systolic images.

No statistically significant differences were observed in LV functional parameters derived from semiautomatic contour detection algorithm as compared with manual contouring, with excellent agreement (concordance-correlation coefficient >or=0.80). The semiautomated contour detection algorithm overestimated LV mass (mean difference 12.84+/-3.39) (P<0.001). The evaluated software allowed to quantify LV parameters in a significantly shorter period of time (mean difference 201.68+/-121.98 s, P<0.001). Contrast-to-noise ratio in end-systole and end-diastole statistically differed (mean difference 5.29+/-2.94 Hounsfield units; P<0.01) but did not significantly hamper the semiautomated segmentation.

Objective quantification of LV volumes using the evaluated semiautomated segmentation software is precise, reproducible, and time effective.

CITATION  J Thorac Imaging. 2009 May;24(2):103-9