MRI Multi Planar Reconstruction a possibility?

Will multi planar reconstruction in MRI ever be possible?
What are the main challenges here?

@shubham4mufc that’s a really interesting question, do you work with MPR?

1 Like

It’s already possible, and equally as straightforward as CT, so long as your images are of sufficient resolution in all three spatial directions. You just use the same process as you do for CT in your DICOM viewer or PACS/RIS software.

The main reason it’s not used very much is that MRI scans are often acquired at relatively low resolution in the slice direction, for either TR, SNR, and/or imaging time reasons. It’s often quicker/preferable just to do three scans (axial, saggital, coronal) rather than an isotropic 3D scan, especially since oblique views are rarely used.


That makes a lot of sense @Tim_Rosenow, I was wondering if you have come across MPR being applied in 3D MRI scans - in the isotropic scenario? I’m trying to get my head around why this might provide benefit over Fourier Transform recon.

1 Like

Oh, you’re referring to non-Fourier encoded MRI. In that case, there are some specific cases where this is useful, specifically for functional MRI or acquisitions where adaptive sequences can be used. Here is a reasonable paper with some background, albeit a bit older:


Isotropic 3D sequences have been in regular use since the early 2000 in clinical scanners
a 3D scan uses fourier transformations in 3 directions.
The excitation pulse is applied to a whole volume selectively or non selectively.
The “slice” direction encoding is performed by one looping set of discrete pahse encoding one step of phase encode value per required slice. The in plane phase encoding is performed in thesame way as a 2D MRI looping one step per step in the phase direction acquisition matrix, and frequency encoding is applied during the signal recall (Echo) and readout.
typically this started with gradient echo sequences, then inversion prepared turbo GRE, and now in to TSE variants and other complex forms.
Scan time -= TR x Nphase(slice) x Nphase(inplane) x NEX
Things can be sped up by employing trains of spin or gradient echoes with different phase encoding in plane.
Sped up by applying parallel imaging techniques along one or both of the phase encode directions.
Sped up by reducing acquired phase steps and interpolating the intervening data.
And sped up by applying compressed SENSE methods

1 Like
Category icons made by Freepik from