Retraction to rebound: DIANA paper

Much controversy has emerged since the first appearance of the DIANA paper in the popular literature. Due to the high status of the journal and the high stakes of the claims, a flurry of activities followed to replicate the results which turned out to be elusive. Now, in a nod to the first scientific rebuttal of the paper (Phi Van 2024, A different interpretation of the DIANA fMRI signal), Prof. J.-Y. Park put out a pre-print discussing possibility of contamination of the DIANA signal by artifacts in spoiled gradient-echo imaging. The team, now with a new, rising graduate student (J.-Y. Keum) in the lead, observed non-biological signals akin to DIANA under certain conditions, and concluded that such conditions were likely present in the original experiments. See: Update on the reproduction and interpretation of DIANA fMRI. Given this revelation, the team decided to retract the paper from the journal Science while investigating the exact role that such artifacts played in the initial results. Does this mean the end of DIANA? Unlikely. First, the original paper triggered massive interest in line scan-based fast fMRI which will likely continue in some form. Second, the sheer amount of positive data on detailed association between the DIANA signal and neuronal activities (Keum 2024, Direct imaging of neural activity reveals neural circuits via spatiotemporal activation mapping) cannot possibly be all coincidental. We may still be chasing the signal mechanism. Practically, at any rate, this development underlined the difficulty of separating out a neuronal signal from non-biological artifacts when the former has a low SNR on the order of 0.1%.

MRI research lab featured in public blog

The MRI research lab in NY is featured in the company's public blog, highlighting the lab's history of innovation and recent neuroimaging scanner projects. See here.

In a separate news, the King's College in London, UK has announced the 4th High-Performance Neuroimaging Symposium to be held in December (link). This two-day meeting will prominently feature the Lab's 3T head-only scanner development and applications.

Connecting with former students

 

Former members of the SKKU MRI laboratory had a great time in a get-together dinner meeting during my Korea visit in May. Shown in the picture are (from the right) Byung-Pan Song, who will be married in June 2026; Hyeong-Seop Kim (with V), who is working in the healthcare industry, and Jun-Ho Kim who is cheerfully preparing for life after school. I am also working on a presentation for Kibaek Kim, who was a former undergrad student at GBME and who is now organizing a special seminar for his current school he is graduating from. My best wishes to all!

Much happened in Honolulu too.

I delivered three talks on Monday, Wednesday, and Thursday. The Wednesday talk was as a substitute speaker for Prof. Jang-Yeon Park, on the basics of MRI, in particular on the k-space. The talks are all recorded for later views for the meeting registrants. For example, the educational talk is at https://ismrm2025.blazestreaming.com/sessions/ismrm-2025-w-04 (Starting at 40 min mark).

Thanks to the remote registration option, even on-site attendees can now watch most oral sessions from the hotel room, with the benefit of unobstructed, high-resolution view of all the slides. So if one is not particularly interested in seeing the speaker, nor in asking live questions, this was a handsome option not only for convenience but also for quick hopping among many parallel sessions.

One caveat for my future reference: Uploaded talk slides may have automatic slide advance turned on -- make sure this be turned off. 

On the exhibition floor, Chinese companies are clearly gaining ground, with one company displaying an impressive model of rotating whole-body 1.5 T scanner for multi-orientation scans. Notably, Hyperfine missed the show, presumably for financial reasons. To all who I met in person in Honolulu this week, thank you for your kind greetings, and my best wishes!

Much happened before Honolulu

In a busy, dynamic month, we now have the compact 3T scanner working again in Mayo Clinic, and have obtained the first ever -- in history of MRI -- images out of the low-cryogen human head 7T scanner. The former is significant because of the near-continuous operation of the world's only low-cryogen 3T high-performance head scanner for nearly 10 years. The project was meant to last for about 5 years at the beginning. This feels like a Mars rover outliving its planned operation time. The second, compact 7T images, will be much discussed in the upcoming ISMRM meeting in Honolulu. I wish the scanner were a bit more compact physically, but both compact 3T and 7T systems show the direction of MRI that I think is right for its future. That is, to maintain its unique, non-invasive high-resolution imaging, while reducing the footprint and operation cost in order for better access and new applications. Let there be no peers and don't compete with low resolution imaging devices.

I had travelled to Mayo Clinic for a week last week, to see a good weather there in a very busy schedule. Mayo Clinic is thriving, with many expansion constructions on display on the streets.

11.7T in Korea in superconductor science & technology journal

Well, so it was published, finally, last month (SUST article, Feb 2025). A few facts according to the paper: The magnet warm bore is 70 cm in diameter, the magnet runs in persistent mode and has been stable in the last 2+ years. In comparison with the Iseult magnet in France, passive shielding (requiring massive shielding walls) and persistent mode (as opposed to driven mode) appear to be the main differences in addition to the magnet bore size (90 cm for Iseult). I could not find gradient coil specification, but from my visit to Gachon a few years ago, the coil's small thickness is unlikely to allow windings with high gradient amplitudes. The paper included a 3D gradient-echo image of an ex-vivo monkey brain with 12.8 cm FOV and 0.5 mm isotropic resolution. This shows the imaging system is operational but the authors certainly would want more images to meet the magnet's high expectation. So, congratulations to the team, and let's hope more is coming.

Compact 3T service visit to Mayo Clinic

February 2025 marks the 9th anniversary of the delivery of Compact 3T scanner to Mayo Clinic. This year, the week of February 17th was a frigid one, not best for air travel, but a service visit to Rochester, Minnesota was made to replace and renew the magnet's cooling system. Thanks to the hard work of the GE HealthCare colleagues and help from the Mayo Clinic support team, the week's task was completed in time. The work will continue until the end of March, with more service visits to ramp the magnet and calibrate the scanner. Hopefully the refreshed 3T system, still the world's only high-performance compact 3T MRI scanner for humans, will serve the research community for many years to come. 

ISMRM 2025 Abstract Announcement

In my counting, 35 x 151 + 23 = 5308 abstracts are accepted, of which 851 (16%) are oral, and 549 are power pitches. Two submissions of mine were both accepted as an oral presentation, which is not a common event based on the probability (0.16^2 = 2.6%). In addition, Lydia's implanted lead heating work, from Mayo Clinic and where I am the second author, was also accepted as an oral. So apparently you don't have to do low-field or machine learning to get invited to speak in the conference. Now whether I will actually fly to the conference and present in person is another question. I wish remote presentation option was still available after the pandemic.