4D Flow MRI MATLAB Toolbox

There is evidence that in order to improve medical diagnosis it is necessary to implement quantitative methodologies that provide diagnostic support and reduce the probability of diagnostic errors (1). However, in 2022, a paper demonstrated the variability that exists in four commercial software used to evaluate 4D flow MR images, based on data from different vendors (2). This is why there is a need to develop software that is accessible to both the scientific and clinical communities, facilitating integration with new developments in a simple and fast way. It is important that as a community we know what algorithms are running in these software packages, and that it is not just a black box.

In this work we present a toolbox developed in MATLAB (MathWorks, Natick, MA, USA), which allows both the scientific and clinical community to process 4D Flow MRI data in a fast and friendly way. A series of hemodynamic and geometric parameters of large blood vessels can be obtained from this toolbox (Table 1). 

To process the 4D Flow MRI data, we apply image offset and noise correction (3), then generate an angiographic image from the 4D Flow MRI velocity data (4), together with segmentation of the region of interest by a semi-automatic process. If aliasing correction is required, it can be corrected using the Laplacian algorithm (5). From the segmentation a tetrahedral finite element mesh is generated using the MATLAB toolbox iso2mesh (6). Next, the velocity data from the 4D Flow MRI images are transferred to the nodes of the finite element mesh. Subsequently, the hemodynamic and geometric parameters shown in Table 1 are calculated. For data saving, the toolbox allows storing these data in different formats, MATLAB, EXCEL (Microsoft Corporation), VTI and VTK (Visualization toolkit).

We also integrate advanced algorithms to estimate pressure from 4D Flow MRI data processing as vWERP (7) and SAW (8). We have begun to incorporate algorithms such as 4DFlowNet (9), to obtain super-resolution of 4D Flow MRI images. Additionally, we have also integrated a manual multiplanar reformatting module, which allows the extraction of 2D Flow slices from the 4D Flow MRI data. These 2D Flow slices are then saved in a SEGMENT (MEDVISO) software compatible format to estimate a series of flow parameters.

Figure 1 shows the main window of the toolbox, with the segmentation of the aorta for the 3 orthogonal views, with the respective velocity values inside the geometry. Figure 2 shows the module that generates the tetrahedral finite element mesh and where the velocity information is transferred using different interpolation techniques. Figure 3 shows the multiplanar reformatting window, where we can extract a 2D Flow slice from the 4D Flow MRI.

This development has made it possible to characterize these hemodynamic parameters in different groups of patients. For example, patients with bicuspid aortic valve (10), aortic dissection (11), transposition of the great arteries (12), among others. It is expected in the short term to incorporate in this development an intracardiac 4D Flow MRI analysis module, together with new artificial intelligence tools such as: automatic segmentation of large vessels, image noise reduction, fully automatic anti-aliasing techniques, and automatic multiplanar reformatting.

Open-source software access instructions:

To access the toolbox, you can download using this GitHub repository (https://github.com/JulioSoteloParraguez/4D-Flow-Matlab-Toolbox).