Stereological estimation of left-ventricular volumetric and functional parameters from multidetector-row computed tomography data.

This study aims to optimize the stereological method for estimating left-ventricular (LV) parameters from retrospectively electrocardiography-gated 16-row MDCT and to compare stereological estimations with those by MRI. MDCT was performed in 17 consecutive patients with known or suspected coronary disease. Stereological measurements based on point counting were optimized by determining the appropriate distance between grid points. LV parameters were evaluated by standard CT analysis using a semi-automatic segmentation method. Two independent observers evaluated the reproducibility of the stereological method. End-diastolic volume (EDV) and end-systolic volume (ESV) estimations with a coefficient of error below 5% were obtained in a mean time of 2.3 +/- 0.5 min with a point spacing of 25 and 15 pixels, respectively. The intra- and interobserver variability for estimating LV parameters was 2.6-4.4 and 4.9-8.2%, respectively. MRI estimations were highly correlated with those by standard CT analysis (R > 0.82) and stereology (R > 0.84). Stereological method significantly overestimated EDV and ESV compared to MRI (EDV: P = 0.0011; ESV: P = 0.0013), whereas for stroke volume (SV) and ejection fraction (EF), no difference was observed (P > 0.05). For standard CT analysis and MRI, significant differences were found except for SV and EF (EDV: P = 0.0008; ESV: P = 0.0004; EF: P = 0.051; SV: P = 0.064). The time-efficient optimized stereological method enables the reproducible evaluation of LV function from MDCT.