A normal placenta development is crucial for a successful pregnancy. Placental function depends on its vasculature pattern such as capillary structure, spatial arrangement, blood flow rate and capillary numbers. Abnormal uteroplacental vascurlarization can cause major obstetric complications such as intra-uterine growth restriction (IUGR) or abnormally invasive placenta (AIP). Diagnosis accuracy of AIP is subjective depending on trainings and experiences of operators. Clinically, 3D ultrasound and Doppler imaging are used to understand utero-placental morphology and flow dynamics. However, maternal and fetal flow in utero-placental unit cannot be discriminated, blocking a better understanding of the pathology to be achieved [1]. A comprehensive and quantitative three-dimensional understanding of the vascularization is still lacking even ex-vivo. Conventional ex-vivo quantification and 3D description of placental vascularization relies on series of 2D histological slices with samples dissected from the placenta. Micro-CT can be applied on small animal placenta fixed in formalin phosphate [2], but with a long and complicated preparation procedure. Magnetic resonance angiography (MRA) may be used to complement such concerns. In this work, we presented the flexibility of placenta vascularization quantification on a fresh healthy human placenta using ex-vivo MRA. A quantification framework is presented with a set of systematic metrics defined to characterize the vascularization.

Material and MR imaging (Fig. 1): A fresh human placenta, from a normal pregnancy, was obtained immediately after delivery. It was catheterized from the top umbilical vessels. The placenta sample was cleaned with the residual blood completely washed out. Pump oil (Vacuubrand®) was chosen as a contrast agent in order to provide optimal contrast with minimal extravascular diffusion. The perfusion liquid quantity/placental volume ratio was approximately 1:1. The MRA scan was performed on a 3T clinical scanner (GE) using a 3D FGRE sequence. The relevant imaging parameters are: field of view 20 cm, NEX 10, acquisition matrix 440×440×116, in-plane resolution 0.5mm, slice thickness 0.6 mm, TR/TE 9.44/2.38 ms. The overall acquisition time is 80 min. The work is approved by the local ethic committee.

Data analysis (Fig. 1): The dataset intensity was first enhanced in order to highlight the arteries. The vessel network was then segmented by an adaptive thresholding method. The voxel 3D skeleton was extracted using a 3D medial axial thinning algorithm [3]. This skeleton was then converted into a network topology described by nodes and connections [4] to understand its branching pattern. Morphological metrics (defined as below) were then calculated based on the extracted topology. All the processing and computation were performed using a home-made software implemented in MATLAB.

Vascularization metrics: morphological indices in clinical ultrasound exams are applied here. We also borrowed quantitative metrics used in tissue engineering to evaluate the tissue engineered micro-vascularization [5] to describe the growth of the vascularization network. Vascularization volume is the volumetric sum of all the detected MRA enhanced vessels; average diameter is defined as the average of all the detected vessels ; 3D vessel length is defined as the total curvilinear length of the vessel tree skeleton; average tortuosity is defined as the average of the ratio of curvilinear distance and Euclidean distance between the starting and the end of a vessel path [2] (Fig. 2b).

The acquired data volume is almost isotropic. The placenta artery network can be clearly visualized (Fig. 2). A 3D model of the vessel tree architecture is rendered in Fig. 2a. The model is color-coded according to the vessel diameter. The network topology is represented as 3D skeleton and plotted in red (Fig.2b). The overall volume of the vascularization is 46.5 ml. The mean diameter of the vessel is: 0.75mm (minimum 0.30 mm, maximum 6.44 mm). The average tortuosity of each vessel path is 1.96. The 3D skeleton length is 3.5×10³ mm. We proposed a framework to quantify the vascularization of placenta on an ex-vivo MRA dataset of fresh human placenta perfused by pump oil. A set of comprehensive description of the vascularization is also proposed. The minimum diameter that the current protocol can detect is 0.3 mm, which is around the protocol resolution limit. The future work is to recruit more volunteers to compare the morphological vascularization indices between physiological and pathological placenta. In the end, we hope to apply the quantification method to the diagnosis of abnormally invasive placenta once the in-vivo MRA protocol is successfully established.