Sometimes we run across retraction notices that are vague, and others that are contorted, but we’ve just found one that gets highest marks for being completely inscrutable.
The article, “Bayes Clustering and Structural Support Vector Machines for Segmentation of Carotid Artery Plaques in Multicontrast MRI,” was written by a group from China and Cambridge University in England — so, we’re thinking language ought not to have been much of a barrier to clear English. It appeared in November 2012 in Computational and Mathematical Methods in Medicine, and describes a way to analyze carotid artery plaque levels in MRI images.
But according to the notice, the technique did not work as planned (or so we think):
This article has been retracted upon the authors request as it was found to include unreliable interpretation due to insufficient provision of studying materials.
Frankly, we have no idea what that means. Were the researchers unable to accurately analyze their results because they didn’t have enough samples? Did someone fail to give them adequate samples in the first place? (In either case, it’s a mystery as to how could they do the study in the first place without sufficient provision of studying materials.) Did someone ask for their materials and they couldn’t provide them, therefore calling the entire study into question? And what are “studying materials”, anyway? Materials to study? Or equipment with which one studies samples?
The whole thing puts us in mind of this exchange from “When Harry Met Sally“:
Harry: [about Auld Lang Syne] What does this song mean? My whole life, I don’t know what this song means. I mean, ‘Should old acquaintance be forgot’? Does that mean that we should forget old acquaintances, or does it mean if we happened to forget them, we should remember them, which is not possible because we already forgot?
Sally: Well, maybe it just means that we should remember that we forgot them or something. Anyway, it’s about old friends.
We’ve emailed one of the authors, Jonathan Gillard, of Cambridge, for an assist and will update this post if we hear back.
The article has only 4 citations, according to Google Scholar.
Like Retraction Watch? Consider supporting our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post.
Just judging from the title of the paper and the name of the journal, I think it’s a good bet that “studying materials” is meant to refer to data with which to train an adaptive algorithm; perhaps “multicontrast MRI” of “carotid artery plaques” that have already been (correctly or at least credibly) “segment”ed by some other method (perhaps inspection by human experts). At least that’s my prior. Somehow I doubt that, if I do look at the article, I’ll have the expertise to form an opinion that’s likely to budge it much. But I’ll try later anyway (unless someone who actually knows something comments relevantly first).
I think it is because the method was tested on only one image. Hard to conclude anything from that. It should have been tested on a much larger dataset.