The workshop was held in a small auditorium/lecture room from Oct 12-14, with remote listen-in allowed for paid registrants. Remote attendance option in many such occasions is a notable positive change since the pandemic. The morning session of the second day covered MRI systems engineering, providing opportunity to hear about latest development or lack thereof from a few well-known speakers. My summary from the first 4 talks of the session is:
1. The French 11.74T system has identified resistive 3rd order shim coils coupling to the gradient coils as the primary source of image artifacts. They also found that other UHF MRI systems employing the same (Siemens SC72) gradient coil exhibited similar behavior. I am sure this will spur much discussion on the coupling issues, potentially exacerbated by vibration, between various non-RF coils in MRI.
2. Baby MRI researcher at 7T reported successful use of MR-based electrical properties tomography to estimate permittivity of infant brains.
3. Siemens's most powerful head-only gradient coils were apparently designed in Massachusetts General Hospital. The presenter incorrectly said that BEM is the industry standard method of gradient coil design. He also repeated what I think is a misleading statement that their design reduced PNS - peripheral nerve stimulation - in 4 out of 5 scan positions tested. The statement gives an impression that the 5 positions were selected randomly, while I believe they selected lowest-risk ones.
4. Brian Rutt continues to work with Minnesota team to provide a head-only gradient coil to fit inside a whole-body gradient coil that is in Minnesota's 10.5T scanner. He covered a broad range of topics including Comsol-based MGI simulation. He talked about an ultimate head gradient coil (reducing both PNS and MGI) as a work-in-progress.
The first day's schedule included Prof. Jang-Yeon Park's DIANA talk. I missed that one, but hope that the DIANA work will overcome a flurry of ego-driven criticisms (particularly instigated by one sore, petty old man) and be known for its true potential.
PS. The Workshop organizers kindly made recordings of presentations and Q&A available on-line until Nov. 9th. In the recording, Peter Bandettini was not being objective - he was politely but clearly siding with the petty man. It wasn't like him when he said "how can a higher-field be worse?", and when he repeatedly said a real signal will ALWAYS show up as number of averages goes up. It is unbelievable that a person with his expertise said something so obviously non-obvious so confidently. Fortunately two people from the audience brought these points up, and predictably he could not give a clear answer. Something is not right here. In a sense, I saw hope.
The tone of your summary of my talk is disappointing. My sense is that the international community views Dr. Seong-Gi Kim with extremely high regard. His rigor, insight, and expertise are among the very best in the world in what he does. In no way would I consider him petty (assuming you are referring to him in your comments). He cares deeply about the science and wants it to be correct. No one is criticizing DIANA. We are all trying to replicate the results. The international community has embraced data sharing. It would be helpful if Dr. Park offered to share his data (prior to pre-selection and after) so that we could all determine precisely how the selection was done. The primary issue that Professor Park alluded to was that the activation induced signal was not consistent, and that more averaging would result in worse results and not better. As the person in the audience asking a question correctly pointed out after my talk, if the signal starts out strong and then over time, diminishes, certainly more averaging would not help. My "confident" statement was with the qualifier that the signal we are looking for is consistent. If it is not consistent then how is the signal pre-selected? This is a non-trivial step. Dr. Park did not mention how this was done, and this piece of information is absolutely essential. I'm also somewhat confident that higher field usually helps increase the sensitivity. This is not really going out on a limb, but again, I could be wrong here - I welcome discussion. Lastly, I'm not taking sides. This is not about taking sides. This is about reaching a better understanding and trying to figure out how to robustly replicate these exciting results. If Professor Park shared his data, both pre-selection and post-selection, and specified precisely how the selective averaging was performed, it would be deeply helpful to everyone and to the promotion of DIANA.
ReplyDelete