To investigate the feasibility and influencing factors of three-dimensional (3D) fully automated dynamic volumetric quantification technique for the measurement of biventricular volume and ejection fraction.

Fifty-five patients with coronary heart disease confirmed by coronary angiography at the First Hospital of China Medical University from January to December 2020 were examined by 3D fully automated dynamic volumetric quantification technique (left ventricle: DHM; right ventricle: 3D RVauto) and cardiac magnetic resonance (CMR). The left and right ventricular end-diastolic volume (LVEDV and RVEDV), end-systolic volume (LVESV and RVESV), and ejection fraction (LVEF and RVEF) were measured, and manual adjustment (DHME and 3D RVE) was performed for some cases in which endocardial boundary automatic recognition did not conform to the actual boundary. Consistency test was conducted between the measured results and CMR values. The patients were further divided into subgroups according to image quality, heart rate (HR), left ventricular segmental movement abnormalities, left ventricular morphological changes, RVEDV, and RVEF to investigate the accuracy and influencing factors of 3D fully automated dynamic volume quantification technique.

There was no significant difference in LVEF values measured by DHM, DHME, and CMR (P＞0.05). LVEF values measured by DHME showed a highly positive correlation with those measured by DHM (r=0.94 vs 0.80), while the accuracy of DHM in LVEF measurement was within the clinically acceptable range (percentage error＜30%). Consistency between values measured by DHM and CMR was significantly decreased in patients with poor image quality, HR＞63 beats per minute (bpm), segmental wall motion abnormality, and left ventricular morphological changes (percentage error: 24.56% vs 13.84%, 28.01% vs 19.14%, 27.29% vs 18.83%, and 24.15% vs 23.80%, respectively), but all the percentage errors were within the clinically acceptable range. There was no significant difference in RVEF values measured by 3D RVauto and CMR (P=0.06). The correlation of values measured between 3D RVE and CMR increased (r=0.95 vs 0.49). Consistency between values measured by 3D RVauto and CMR was significantly decreased in patients with poor image quality, HR＞63 bpm, increased RVEDV, and decreased RVEF (percentage error: 44.50% vs 32.47%, 39.48% vs 41.07%, 46.24% vs 38.08%, and 41.42% vs 31.16%, respectively), and the consistency was significantly improved after manual adjustment (percentage error: 12.36%, 13.31% and 13.40% and 11.29%, respectively).

3D fully automated dynamic volumetric quantification technique has a high accuracy in measuring left ventricular systolic function, and can be recommended for extensive clinical application. However, its accuracy in measuring RV systolic function is generally low. For patients with poor image quality, HR＞63bpm, segmental wall motion abnormality, left ventricular morphological changes, increased RVEDV, and decreased RVEF, the measurement accuracy can be improved by manual adjustment of endocardial boundary.