Automatic multi-vessel volume flow calculation with 4D flow CMR

Abstract

Volume flow analysis is essential in the assessment of many cardiovascular diseases such as valvular regurgitation, intra-cardiac shunt, and complex congenital heart diseases. Clinically, CMR-based volume flow analysis is performed using 2D flow CMR. This requires user-dependent and time-consuming positioning of 2D planes in each vessel while the patient is still in the scanner. Previous studies have demonstrated that 4D flow CMR permits accurate volume flow assessment. However, retrospective plane-positioning and region-of-interest delineation requires time-consuming user interaction. The aim of this study was to develop an automatic method for volume flow analysis in the great thoracic vessels using 4D flow CMR. Methods The automatic multi-vessel volume flow calculation method is illustrated in Figure 1. An atlas (reference vessel segmentation) was created by manual segmentation of the great thoracic vessels in one healthy volunteer. The segmentation was done on a 3D PC-MRA which was derived from the 4D flow CMR data. Analysis planes for volume flow determination were positioned in the proximal ascending aorta and pulmonary trunk. For each subject, the atlas’ PC-MRA was registered to the subject’s PC-MRA. In this way, the atlas’ vessels and analysis planes were transformed into the subject’s vessels. The transformed atlas was transferred to all timeframes using the 4D flow CMR magnitude image, resulting in a time-resolved segmentation that follows the motion of the vessels over the cardiac cycle. Finally, the volume flow was automatically calculated for each plane using the time-resolved atlas as a mask to account for vessel location, shape and movement. The method was evaluated in a group of subjects composed of 10 healthy volunteers and 11 patients with heart failure of different etiologies. Results in the proximal ascending aorta were compared against volume flow values obtained by manual segmentation. Additionally, the pulmonary-to-aortic flow ratio (Qp/Qs) was assessed. Results