Post-processing in cardiovascular computed tomography: performance of a client server solution versus a stand-alone...

To compare the performance of server-based (CSS) versus stand-alone post-processing software (ES) for the evaluation of cardiovascular CT examinations (cvCT) and to determine the crucial steps.

Data of 40 patients (20 patients for coronary artery evaluation and 20 patients prior to transcatheter aortic valve implantation [TAVI]) were evaluated by 5 radiologists with CSS and ES. Data acquisition was performed using a dual-source 128-row CT unit (SOMATOM Definition Flash, Siemens, Erlangen, Germany) and a 64-row CT unit (Brilliance 64, Philips, Hamburg, Germany). The following workflow was evaluated: Data loading, aorta and coronary segmentation, curved multiplanar reconstruction (cMPR) and 3 D volume rendering technique (3D-VRT), measuring of coronary artery stenosis and planimetry of the aortic annulus. The time requirement and subjective quality for the workflow were evaluated.

The coronary arteries as well as the TAVI data could be evaluated significantly faster with CSS (5.5  ±  2.9  min and 8.2  ±  4.0  min, respectively) than with ES (13.9  ±  5.2  min and 15.2  ±  10.9  min, respectively, p ≤  0.01). Segmentation of the aorta (CSS: 1.9 ±  2.0  min, ES: 3.7  ±  3.3  min), generating cMPR of coronaries (CSS: 0.5  ±  0.2  min, ES: 5.1  ±  2.6  min), aorta and iliac vessels (CSS: 0.5  ±  0.4  min and 0.4  ±  0.4  min, respectively, ES: 1.6  ±  0.7  min and 2.8  ±  3  min, respectively) could be performed significantly faster with CSS than with ES with higher quality of cMPR, measuring of coronary stenosis and 3D-VRT (p < 0.05).

Evaluation of cvCT can be accomplished significantly faster and better with CSS than with ES.  The segmentation remains the most time-consuming workflow step, so optimization of segmentation algorithms could improve performance even further.