HARDI - multiband


Im working on a pediatric connectome project and would like to know if anyone has experience with multiband connectome sequence on Philips.

can you share a protocol for Multi-Shell Multi-Tissue Constrained Spherical Deconvolution? or provide parameters for this sequence?



Good b-value sensitivity parameter for 2shell

The acquisition protocol developed for the dHCP sequence is published in MRM.


While the dHCP protocol is a good starting point, the sequence is so heavily customised that there’s essentially no chance of anyone else being able to replicate without some serious development… It’s also specifically for neonatal imaging, I’m not sure that it would necessarily be optimal for pediatric imaging – I expect for that age range, adult protocols would be pretty close. But the considerations are very different, mostly about the logistics of keeping the kids in the scanner, motion sensitivity, and overall time constraints.

@ameoded1, I expect you’re asking about setting up a protocol using the product sequence? I’d start by figuring out what your time budget is, what resolution you’re targeting, and what multiband factor you can achieve without degrading image quality. That would then dictate how many volumes you can acquire. At that point, you can start thinking about how many shells you can hope to acquire, given that you’ll want a reasonable number of DW directions for each shell, with denser sampling on the higher shells to recoup SNR and properly sample the higher angular frequency content. Finally, the last question is what b-values for each shell, and how many DW directions for each b-value – that one is still an open question, but I reckon it’s likely to be a fairly broad optimum.

I appreciate that’s still pretty vague, but hopefully it’ll be a useful starting point…



  1. Lifespan connectome protocol:
    80 dir, 5 b0, 38 b1000, 37 b2500

  2. dHCP: 90 dir, b1000, 2000, 3000. 6 runs each 9 min aprox. resl 1,25 mm isotropic.

I guess I will have to compromise and come up with a solution suitable for children. I have no more than 30 min-45 min scan time including structural images. b values: 0, 1000, 2000, 3000

I’m not sure what MB factor to choose, I see MB 3 in HCP.

Thank you



Sounds like a different dHCP than ours. These parameters look like the adult HCP protocol… (?).

For our dHCP, we have:

  • 20 × b=0 s/mm²
  • 64 × b=400 s/mm²
  • 88 × b=1,000 s/mm²
  • 128 × b=2,600 s/mm²

at resolution 1.5mm isotropic, multiband factor 4 (full details here).

That depends entirely on how good your multiband implementation is, including the RF DAC & amplifier, the receiver coil array, and the sequence itself. This is something I’d definitely advise you experiment with to figure out the limits of your system…


Perfect. I see that (reference) is optimized for neonatal MRI.

Any chance you can provide reference for multi shell HARDI for young adults?

This is what I found in Stanford website:

DTI g150 b3k 3-shell (6:15): 3-shell with 10 b=0 images, 30 direction at b=1000, 45 direction at b=2000, 65 direction at b=3000. SMS factor 3, axial slices, 2x in-plane acceleration, voxel size 2mm^3, number of muxed slices 21 (63 unmuxed slices, 12.6 cm)



I’m guessing you’re asking for a reference about the optimal settings for multi-shell HARDI in young adults…? In which case, I’m not aware of any specifically. Sorry…