DISCO-mcDESPOT Nov. 6, 2011 Jason Su.

Presentation on theme: "DISCO-mcDESPOT Nov. 6, 2011 Jason Su."— Presentation transcript:

1 DISCO-mcDESPOT Nov. 6, 2011 Jason Su

2 Methods In vivo, 3T Possible issue with collection
2mm in-plane resolution 5mm slice resolution Rect 200us SPGR (13 angles from up to 18) and SSFP (13 angles up to 65) Left out B1 correction Running into issues with HIFI producing completely wrong T1 Still some work to do incorporating Ives’s Bloch-Siegert -> kappa map code into pipeline Possible issue with collection May need more samples on low end of SPGR signal curve, was enough for DESPOT1 but perhaps not mcDESPOT, only 1-2 points before Ernst angle

3 Results Performance is mixed for the different output maps
Overall good performance with T1 Worse performance with T2, probably because bands caused by off-resonance affect high-freq. k-space MWF % difference histogram is poor but looks decent, could be a dividing by small numbers issue Stochastic nature of optimization makes interpretation difficult, i.e. how much variation is due to randomness or view sharing?

4 MWF – offline.recon

5 MWF – DISCO

6 MWF – % Difference Mean: SD:

7 Fast T1 – offline.recon

8 Fast T1 – DISCO

9 Fast T1 – % Difference Mean: SD:

10 Slow T1 – offline.recon

11 Slow T1 – DISCO

12 Slow T1 – % Difference Mean: SD:

13 Fast T2 – offline.recon

14 Fast T2 – DISCO

15 Fast T2 – % Difference Mean: SD:

16 Slow T2 – offline.recon

17 Slow T2 – DISCO

18 Slow T2 – % Difference Mean: SD:

19 Off-Resonance – offline.recon

20 Off-Resonance – DISCO Pattern shift is quite apparent here, I suspect because DISCO is losing a lot of the information that happens at the banding edges

21 Off-Resonance – % Difference
Mean: SD: The histogram is clearly non-Gaussian, more reason to believe that this is a particularly affected map. Large SD is due to large tail on + side not shown in histogram.

22 Residence Time – offline.recon

23 Residence Time – DISCO

24 Residence Time – % Difference
Mean: SD: 16.17

25 More Images From the performance in T1 mapping, we know that our SPGRs are fairly high-fidelity SSFPs are probably the source of error The first flip angle is again problematic In the future, should probably start with improving DESPOT-FM performance, much easier to look at

26 SSFP fa5 – offline.recon

27 SSFP fa5 – DISCO

28 SSFP fa5 – % Difference