ARTICLE -- Scientists Fraudulently Manipulate Digital Images

[mmcclellan]

Journals Find Fakery in Many Images Submitted to Support Research
By JEFFREY R. YOUNG

Kristin Roovers was a postdoctoral fellow at the University of Pennsylvania with a bright career ahead of her—a trusted member of a research laboratory at the medical school studying the role of cell growth in diabetes.

But when an editor of The Journal of Clinical Investigation did a spot-check of one of her images for an article in 2005, Roovers's research proved a little too perfect.

The image had dark bands on it, supposedly showing different proteins in different conditions. "As we looked at it, we realized the person had cut and pasted the exact same bands" over and over again, says Ushma S. Neill, the journal's executive editor. In some cases a copied part of the image had been flipped or reversed to make it look like a new finding. "The closer we took a look, the more we were convinced that the data had been fabricated or manipulated in order to support the conclusions."

[...]

Read on at--

Dead Link Removed

--------------------------------------------------------------------------------
Copyright © 2008 by The Chronicle of Higher Education

 

[df cardwell]

Well, gosh.

Not quite the old romance about the young scholar facing temptation
in a far off library, stealing an unknown letter which refutes a doctoral thesis !
(Dorothy L. Sayers fans, g'day). But giving Photoshop to researchers is like giving
handguns to customers at a liquor store.

Much of the reason I despise burning and dodging a print is that I was forbidden those manipulations when I was a darkroom technician for a Pathology Lab in my college years. I was allowed to fit the scale of the negative to the paper by contrast grade and developer, and each run of prints had to have descriptive notes to accompany the pictures. Old School.

Digital imaging is wonderful for this sort of scientific work,
being able to accurately present what you can see with your eye,
but unless the operator has acquired character along with an advanced degree,
it is all pointless. It should be pretty easy to decide what adjustments are allowed,
which are not. Treat the imaging steps as any other protocol, and require digital certification
of how the image was processed. Not hard, really.

Thanks Michael.

 

[Lopaka]

Treat the imaging steps as any other protocol, and require digital certification
of how the image was processed. Not hard, really.

Exactly. There are specific protocols in forensic photography which evidence technicians must follow.

Bob

 

[Antje]

Interesting... We had such a case at the research center I got my PhD at. The girl was fired, of course. She also faked protein bands - but she actually scanned them from the catalogue of a lab equipment vendor. Meaning on top of being a total fake and cheat, she didn't even have to run the experiment to get those "marvelous" results.

Still - it's not Photoshop that's the culprit. I'm a biologist, and I spent many many desperate nights at the lab, not being able to find the damn cell growth I was hoping for. It would have been soooo easy to just sprinkle a little interleukin-2 on my cells, randomizing the amount I put on them to cover my tracks, and shove them into the cell counter. Nobody would have been able to tell - and I even could have made pictures of the cell growth on slide film! I could even have shown that miracle to anyone interested live and in person. There would have been no way of telling what I would have done as cell activation via the IL-2 receptor was what I was looking at. I would have manipulated my experiment at the most basic level. I didn't, I never saw the activation I was looking at, and later found that the principle I was studying didn't even allow for cell activation like that - everything was totally right after all, and the missing cell growth pointed me into the right direction in my research.

But it would have been so easy... and it would have made my boss happy... Oh, and the output of the gamma counter was printed on just plain white paper, the data weren't stored anywhere, how could anyone have told I didn't just print them using my office printer? Apart from the simple fact that it would have been just as easy to fake a nice experiment just by putting isotopes into the wells instead of the cells I was studying. And histology - faking cell staining would have been laughably easy and would have created durable slides.

The point is: Science doesn't work without honesty and trust. Photoshop or not. The instances where you could actually use Photoshop to manipulate have been few in my career at least. My work was measured in plain numbers and microscopic slides mostly...

Antje

 

[winger]

Having worked in a forensic lab for 14 years, photos of evidence there do not get doctored. In a lab like that, there's no motive to do so. In research, the whole publish or perish atmosphere contributes greatly to the desire to show perfect results.

 

[Antje]

Exactly. There are specific protocols in forensic photography which evidence technicians must follow.

Bob

As long as you don't manipulate what you want to analyze.

Antje

 

[Antje]

In research, the whole publish or perish atmosphere contributes greatly to the desire to show perfect results.

I couldn't agree more.

Antje

 

[David A. Goldfarb]

I've posted a link to the full text of the article in the initial post, following the opening paragraphs, which tell what the article is about. Please don't post the full text of an article published elsewhere under copyright on APUG.

 

[wfe]

Neither Photoshop nor technology are the culprits here, people are.....

~Bill

 

[nc5p]

Science is becoming more and more like this. The dangerous part is that national and local governments routinely pass laws based on this type of hype. They also endanger lives by using this junk science in medical treatments.

 

[removed account4]

heinrich schliemann, the father of modern archaeology, fudged his data at troy from what i remember.
faking results is not new, and has been around for as long as people have wanted to take that route ...
what was that saying again .....
oh yeah .. believe none of what you hear and half of what you see..

 

[Struan Gray]

I actually think the problem was much worse in the days when people would hand-draw figures.

Linda Miller makes the most intelligent comment in that article: data with noise is more believable than data without. Noise is much more difficult to fake well than signal

In my own work I don't check every picture I see in every journal. I do check very carefully any images that are related to my own research, or which are used to support conclusions that affect me directly. When I referee papers I usually put images through a torture test to see if I can detect heavy-handed 'clarification'. This is actually a lot easier to to now that most research is publised electronically.

There is so much published these days that it would be very surprising if there were no fraud. The days when you would spend three or four years working on a problem and then publish a synthesis of the whole work are long gone. Now, every last intermediate datapoint is published, and becuase people are expected to publish complete papers, they are usually accompanied by a useless pile of speculation masquerading as analysis. There is no requirement for your data to make systematic sense: it just has to exist, and preferably be a little informative or intriguing. Scientists are just following the money, if you want someone to blame, blame the politicians who think they can measure creative output as it happens.

 

[firecracker]

Journals Find Fakery in Many Images Submitted to Support Research
By JEFFREY R. YOUNG

Kristin Roovers was a postdoctoral fellow at the University of Pennsylvania with a bright career...

I suppose she could've walked over to the Annenberg Center/School and sat there to get a lecture from someone on the ethics on digital imaging, etc...

 

[jpeets]

They also endanger lives by using this junk science in medical treatments.

What are you referring to?

The review of data by health authorities is comprehensive and is based on a large number of studies that are audited. This article has to do with a single basic research study, self-reported by the investigator. No relevance to approval of medical treatments.

 

[Antje]

Linda Miller makes the most intelligent comment in that article: data with noise is more believable than data without. Noise is much more difficult to fake well than signal

So true. That's why I despise those fraudsters even more: Not only did they cheat, they also cheated badly. Not a sign of extreme intelligence... Well, at least they're easier to find that way.

Hey, did you hear the story about the lab tech who faked a study about cancer and mobile phones? She "showed" that mobile phones cause DNA strand breaks in cell culture. Those studies are the main ammo in most German anti-mobile phone discussions, and they're just plain faked. Apparently, nobody checked her data. That's something I experienced a lot when the bosses liked the results they see.

Not that my boss ever liked my results, and maybe that's a good thing, as it forced me to work very hard to convince him... Actually, I found out after two years of work and myriads of red and green immunoflurescence studies that my boss is red-green blind. Really. I thought he was just mean when he said he couldn't see the difference, when it was totally clear to everyone else. But I digress.

Antje

 

[Alex Hawley]

faking results is not new, and has been around for as long as people have wanted to take that route ...

That's very true but now its just so much easier to do than its ever been. I think this is going to be an ethical dilemma of huge proportions and the consequences of it can be grave indeed.

 

[Kino]

"...Actually, I found out after two years of work and myriads of red and green immunoflurescence studies that my boss is red-green blind. Really. I thought he was just mean when he said he couldn't see the difference, when it was totally clear to everyone else..."

Where'd you hide the body?

 

[Tom Reardon]

Fraud happens, but it's a relativaly small problem. A larger concern is that many investigators are all to ready to accept results from the computer that have simple data-entry and/or programming errors when a close scrunity of results and the data should have sent up a red flag - I see a fair amount of that in my day job.

 

[Struan Gray]

Actually, I found out after two years of work and myriads of red and green immunoflurescence studies that my boss is red-green blind. Really. I thought he was just mean when he said he couldn't see the difference, when it was totally clear to everyone else.

And he didn't know?

Everyone I have worked for has either had a breakdown, or run away to another country. Let me know if you want to collaborate

 

[Dan Fromm]

Y'know, there are also mistakes. Some years back a young ichthyologist published a phylogeny of the euteleosts that was strongly inconsistent with prior work. A couple of the prior workers were very surprised, asked the guy for his data, redid the analysis, got the same results. Since they didn't believe the results, they then checked his data. Got the specimens, remeasured, ... They found numerous data entry errors. With correct data, the surprising result went away and the conventional one stood up.

Some years ago one of my friends' students included a surprising mtDNA-based phylogeny in her dissertation. Published it, too. I was, um, surprised by one of the data points, went to Panama, went to the location where she'd collected the surprise, found what I thought was it -- mine matched a specimen, unfortunately not the one sequenced, she'd deposited in USNM -- and found she'd misidentified it. Thing was, my mtDNA sequence was not at all like hers. Eventually I did a BLAST search for sequences like the surprise in the NIH database. The first 50 or so matches, after the surprise, were all human.

Honest people make mistakes, don't always catch them. I am, though, still a little mad at my friend for not making the kid compare the surprise with a reference specimen of the fish she thought it was; that's the first thing I did, and it didn't match at all.

 

[df cardwell]

"Honest people make mistakes"

Exactly so, Dan. But that's why we double check the receipt for that S Ortho Planar, why we look over the bill for all those margaritas, and why there is peer review. But 'data fudging' is a weed, and if there is anything notable about this case it is that Post Modernism has convinced scientists that images have no value beyond being illustrative accessories to dress up their words: "There are so many images that a single image has no meaning". She didn't think there was anything wrong with faking the pictures !

Of course, 'the image' has been taking a beating ever since the Enlightenment's apotheosis of The Text to godlike stature.

Sentence the fraudulent writer to a high school biology class, give her a brick of film and a student microscope, and make her stay there until she can hang an exhibition of paramecium.

 

[Monophoto]

I am the editor of a peer-reviewed technical journal, one of many journals published by a large technical organization. Once each year, the editors of all of our journals have a meeting. One of the facts we saw in our meeting a couple of months ago is that the number of instances of plagiarism has doubled every year for the past five years.

My observation on this is that there are two major factors that lead to this kind of behavior. One is where the author (creator) is well-intentioned but naive, and doesn't have a clear grasp of what is OK to do and what is not. But a more insidious problem is where there is a cultural bias that says that it's OK to steal, and that shame is a consequence only when one is caught.

 

[mark]

My observation on this is that there are two major factors that lead to this kind of behavior. One is where the author (creator) is well-intentioned but naive, and doesn't have a clear grasp of what is OK to do and what is not. But a more insidious problem is where there is a cultural bias that says that it's OK to steal, and that shame is a consequence only when one is caught.

Here in lies my problem, Naive is not an excuse. When one is at the level of consideration for inclusion in a peer reviewed journal there is no excuse for not knowing what is and is NOT plagiarism.

Those others you mention should get banned from publishing and lose their positions. Culture has nothing to do with it.

 

[Antje]

And he didn't know?

Everyone I have worked for has either had a breakdown, or run away to another country. Let me know if you want to collaborate

He sure knew, but probably thought it wasn't so bad. :rolleyes: We switched to immunohistochemistry (blue and brown, *really* hard to see), and I did everything all over again. And yes, I thought about killing him, a lot, actually

Hey, I thought Sweden wasn't that bad? Some of our customers work at Karolinska and still seem quite normal. I actually quit science and work for Siemens now... Similar level of insanity, but the pay is much better.

Antje

 

[KenS]

About 6 years ago I published a ‘technique’ paper on the benefits of cross-polarised transmitted light of a zymogram with a new stain that provided a colour the equivalent of a Wratten #8 on the phytase embedded in the gel. Since normal transmitted light (with filtration) could not provide enough contrast to readily observe some of the smaller bands, and could not provide the required photographic/visual contrast with both B/W and Ektachrome. Success was achieved using polarized filtration under the gel… with cross polarization on the lens.

Within the first year of publication, I received a few emails… a few with a ‘great technique’, three emails and one phone call accusing me of “Photoshop Phakery”.

I do not, however have any record of anyone successfully using ‘my’ improved/modified technique.

Ken