The processing of time-resolved 3D phase-contrast MRI with three-directional velocity encoding (4D flow MRI) cases can be highly time consuming given the large multi-dimensional datasets (3D+time of the cardiac cycle+3-directional blood flow velocities). Recent studies have shown that the analysis of aortic 4D flow MRI data can be performed using automated and semi-automated analysis for the extraction of aortic diameters¹, flow quantification^1,2,3, and flow pattern visualization^2,3. However, the fully automated processing of cases in large databases is still challenging. Therefore, the purpose of this study was to introduce an automated workflow allowing the unattended retrospective processing of aortic 4D flow MRI data from a large database of subjects. The specific aims were to: 1) measure maximal aortic diameter (AoD) and peak velocity (PV); 2) provide a quick visualization of 3D hemodynamics; and 3) provide an overview of demographic data and measured parameters.Subjects were identified via an IRB-approved retrospective chart review from a database of healthy subjects and patients who underwent thoracic MRI including 4D flow MRI of the aorta between 2012 and 2015. Inclusion criteria for unattended processing included cases with: a) a complete 4D flow MRI acquisition; b) full pre-processing (eddy-current correction, flow aliasing, calculation of 3D phase contrast angiography [3D PC-MRA]) of 4D flow datasets⁵; c) existing 3D segmentation of thoracic aorta (Mimics, Materialise, Leuven, Belgium) obtained from 3D PC-MRA¹. A total of 782 subjects were included, subdivided into the following groups: healthy control subjects (n=106), patients with bicuspid aortic valve (BAV, n=375), and patients with tricuspid aortic valve (TAV, n=301). All subjects underwent 4D flow MRI during free breathing with adaptive respiratory navigator gating with full 3D coverage of the thoracic aorta4 using 1.5T and 3T systems (Magnetom, Avanto, Aera, or Skyra, Siemens, Germany). Unattended processing was implemented in Matlab (The Mathworks, Natick, MA, USA) as schematically illustrated in figure 1. For each subject, the data analysis workflow included: i) the generation of a velocity maximum intensity projection (MIP) using the 3D segmentation of the thoracic aorta to mask the 4D flow MRI velocity field; ii) the creation of a time-resolved velocity vector movie and depiction of the velocity vector field at peak systole; iii) the automatic calculation of AoD using multiple analysis planes along a volumetric centerline generated from the full 3D PC-MRA segmentation¹; iv) the collection of 4D flow MRI scan parameters and patient demographics from DICOM header. All MIPs, movies, and aortic measurements were saved for future queries or for report generation. PV was automatically obtained from the velocity MIPs and extracted downstream of the aortic valve.Processing times varied from 4-6 minutes/case. Collected 4D flow MRI acquisition parameters were as follows: 1.5 T scan (n=656) used TE/TR=2.4–2.8/4.8–5.4 ms, flip angle α=7–15°, Venc=1.5 m/s, resolution=1.6-2.5×1.6-2.5×2.2-3.4 and a matrix=192–400×108–116; 3T scan (n=119) used TE/TR=2.3-2.6/4.8-5.1 ms, flip angle α=7–15°, Venc=1.5-2.5 m/s, resolution=1.6-2.3×1.6-2.3×2.2-2.8 and a matrix=160-192×80-116. Scan acquisition parameters were incomplete in n=10(1%) of cases. Height and weight were incomplete in n=259(33%) cases. Subject demographics are summarized in Table 1. Examples of velocity MIPs and flow vectors from each group are presented in figure 2. Based on data from the entire cohort of 782 subjects, correlation analysis demonstrated relationships between age and AoD (R=0.34, P<0.001), age and PV (R=0.03, P=0.414), and AoD and PV (R=0.25, P<0.001). The BAV group showed the most significant correlations among groups (Table 2). One-way ANOVA showed significant (P<0.001) group differences for age, AoD and PV. Inter-group analysis, figure 3, showed that TAV subjects were older than controls and BAV, AoD was larger in BAV and TAV subjects, and PV was higher in BAV subjects.This study demonstrated that: 1) unattended processing of 4D flow MRI datasets in a database is feasible; 2) quick visualization of flow patterns was possible with the use of velocity MIPs and vector flow screenshots; 3) group overview analysis can be performed by querying pre-processed cases. It is important to notice that pre-segmentation of the thoracic aorta was needed to generated an adequate visualization of flow velocities and for the calculation of AoD. The automated calculation of PV and flow measurements (net flow, mean flow, retrograde flow) may also streamline the proposed workflow.The proposed workflow allowed for the unattended processing of 4D flow MRI datasets, including the measurement of basic parameters (AoD and PV) and flow visualization (velocity MIPs and 3D flow vectors), in a large database. This represents an important step towards the systematic analysis of 4D flow MRI datasets in large clinical studies.