Remember the Melba Toast image? It was part of a 2005 paper by Surgisphere founder Sapan Desai, published as part of his PhD research at the University of Illinois at Chicago.
The photos in this paper contain many unexpected repetitive features, both within as well as between the panels. This appeared to be a severe case of photographic editing. So I reported this paper to the journal and the university.
Unfortunately, the University of Illinois just let me know that they will not investigate this case – because it is too old.
Sapan Desai was an author and main data-provider for the now retracted Lancet and NEJM papers on COVID-19/hydroxychloroquine treatment. Many concerns have been raised on the validity of the large patient data sets provided by his company Surgisphere that formed the basis of both papers. Since the retraction of both high-profile papers, Surgisphere’s website and Desai’s ResearchGate pages have disappeared (links are to archived pages).
All research papers based on Surgisphere’s data as well as all previous endeavors by Desai are now under scrutiny.
As I reported in a previous blog post, I found serious image issues in one of Desai’s PhD papers called “Comparative Morphology of Rodent Vestibular Periphery. II. Cristae Ampullares” (Sapan S. Desai, Hussain Ali, and Anna Lysakowski – Journal of Neurophysiology (2005), doi: 10.1152/jn.00747.2003, PubPeer post)
The paper contained several photos of rodent brain structures that looked a lot like pieces of Melba Toast, but they were even more remarkable because of repetitive features within each of these photos.
Biological tissues such as brain slices usually do not contain such repetitive features, so this is quite unexpected. It is hard to think of any reason that could explain these repeats other than human photographic editing.
My suspicions were confirmed by two other image forensics, as reported in this Buzzfeed News article. Mike Rossner, former editor at the Journal of Cell Biology and scientific image integrity expert, said: “I concur with the allegation that there appear to be numerous small duplicated regions in the photographs.” Daniel Acuna, a computer scientist at Syracuse University specializing in image duplication detection, used stronger words. “I’ve never seen something like this. It’s outrageous,” Acuna told Buzzfeed News.
Because there are multiple authors on the paper, it is not clear who created this image or what might have caused the repetitive features. One might assume that Desai, as first author and PhD student, would have done most of the lab work. But it is up to the university and journal to investigate this case and find out what really happened. So I reported this paper to the journal and to the University of Illinois at Chicago on June 18th.
Last night, the university reported back that they will not investigate this case, because the paper is too old. The University of Illinois’s Integrity Policy states that:
No Complaint shall be heard or reviewed under this Policy as to conduct alleged to have occurred six years or more before the date of receipt of the Complaint except as described in B.3 below.Policy and Procedures on Integrity in Research and Publication – University of Illinois
And so, it seems that the University of Illinois does not care much about photoshopping in scientific papers by their staff or students, in particular if it happened more than six years ago.
And for that, I am handing them the second “This Image Is Fine” Award. Congratulations, University of Illinois! Good to know that scientific integrity is not a very high priority for your university.
You can read more about the “This Image Is Fine” Award in the first installment here.
6 thoughts on “University of Illinois wins second “This Image Is Fine” Award for Melba Toast paper”
This is all a bit scary, especially the reaction of the University of Illinois. Well done on your detective work!
They look good enough to eat!
Reminiscent of Fig 4C in Nature pubmed: 16625199
I co Chair our University’s Ethics Committee. We also have a cut off for investigating such issues (also 6 years), which as you point out leads to the supposition that we do not care enough to want to follow up such things. Part of the reason for a cut off in time is that we have enough current work to keep us busy, but I also understand the frustration this brings to people looking at our process to “right” “wrongs”. I am not trying to defend this policy, in fact I have always wondered why there are time limitations for the prosecutions of crimes in the US as well. As this was a manuscript that was published it might be an issue for the publisher of the journal. Also, if this were part of work submitted for a degree, then that might change the thought process, I do not remember if my University has had a case where the retraction of a degree might be considered for what clearly (as a result of your meticulous work) seems to be fraudulent data. You raise good questions, thank you for keeping people honest, Science require this.
“in fact I have always wondered why there are time limitations for the prosecutions of crimes in the US as well.”
Tu quoque, anyone?
This is a really unfair take. Statutes of limitations exist for a reason (defendant may not retain exculpatory evidence for such a long time). And as the other commenter points out, they are trying to prioritize which cases to investigate with their limited resources. Even if the university did a full investigation and found against Desai, would that make any difference in how Surgisphere is perceived?