Image manipulation and fraud - Science

Science 17 October 2008
Vol. 322. no. 5900, p. 356
DOI: 10.1126/science.322.5900.356
.
SCIENTIFIC MISCONDUCT:
Falsification Charge Highlights Image-Manipulation Standards
.
Gretchen Vogel (*)
.
Controversy continues to plague work from the lab of prominent stem cell researcher Catherine Verfaillie. The University of Minnesota (UM) announced last week that an academic misconduct committee had concluded that Morayma Reyes, while a graduate student in Verfaillie's lab there, "falsified" four data images in figures in a 2001 stem cell article. The committee found that misconduct allegations against Verfaillie were unsubstantiated, but it did criticize her oversight and mentoring of lab personnel. The new charges come a year after questions were raised about the misuse of images in another key stem cell publication from the group (Science, 02 March 2007, p. 1207).
.
Reyes, now an assistant professor of pathology at the University of Washington (UW), Seattle, and Verfaillie, who now heads the Stem Cell Institute at the Catholic University of Leuven in Belgium, both acknowledge that errors were made in the preparation of the 2001 paper. But Verfaillie defends her supervision, and Reyes says that for several of the disputed images she merely globally adjusted the brightness and contrast in data images without any intent to deceive. "These errors were unintentional and were common and accepted practices at the time," Reyes wrote in an e-mail to Science.
.
The paper, published in Blood, claims that stem cells purified from human blood can form precursors of bone, fat, cartilage and muscle cells, as well as the endothelial cells that line blood vessels. At the time, blood stem cells weren't thought to be that versatile. Verfaillie and Reyes say the figure errors do not alter the Blood paper's conclusions, but Verfaillie has asked the journal to retract the paper, calling it "the proper course in this situation."
.
The Blood paper relates to work that the group later published in Nature, reporting that cells from mouse bone marrow could become a wide variety of cell types. Several groups have reported trouble reproducing that paper's results (Science, 09 February 2007, p. 760). Then last year, Nature conducted a re-review of the paper when a journalist at New Scientist questioned whether some data shown were identical to those in another paper. A UM investigation concluded that any duplication was the result of honest error. Nature published several corrections but said that the paper's conclusions were still valid and that Verfaillie continues to stand by the work.
.
CREDITS: REYES ET AL., BLOOD 98, 9 (01 NOVEMBER 2001) © THE AMERICAN SOCIETY OF HEMATOLOGY
.
New Scientist also alerted the university to an apparent duplicated image in the Blood paper (Science, 30 March 2007, p. 1779). The university then convened a new committee, which submitted its final report on 5 September. The school last week stated that the committee found that in four of the seven figures in the Blood paper, "aspects of the figures were altered in such a way that the manipulation misrepresented experimental data and sufficiently altered the original research record to constitute falsification." The committee cited "elimination of bands on blots, altered orientation of bands, introduction of lanes not included in the original figure, and covering objects or image density in certain lanes," the statement says.
.
The university has not released the full report, citing privacy laws, and experts in image analysis say it is hard to determine intentional fraud solely from the original paper. James Hayden, manager of the microscopy core facility at the Wistar Institute in Philadelphia, Pennsylvania, says that to make a clear point, scientists often alter images, sometimes more than they should. Good laboratory practice means all such adjustments should be noted in a paper and copies of the original image files kept, he says. Jerry Sedgewick, head of the Biomedical Image Processing Lab at UM and one of Reyes's mentors, says he is not convinced that she did anything wrong with the image adjustments she made. "This is done routinely and has been done since film and imaging began," he says.
.
During the investigation, Reyes asked George Reis, who heads the consulting firm Imaging Forensics in Fountain Valley, California, to assess whether changes made between the original image scans and the published images could be due to "global" adjustments, which would imply there was no intent to deceive. Reis told Science that he did determine that significant global adjustments could account for "most of the changes in most of the images." But he says he did not examine the images specifically for signs of editing such as adding or deleting individual lanes.
.
UM says it has forwarded the panel's report and supporting materials to the federal Office of Research Integrity in Rockville, Maryland. UW is waiting for more information from UM before deciding whether to discipline Reyes, according to a spokesperson.
.
Both Verfaillie and Reyes say they have implemented much stricter rules for dealing with data images in their labs as a result of the case. "I have learned a hard lesson," Reyes e-mailed Science. "Now that I am a mentor … I will make sure that my students will get the proper training, supervision and education."
.
*: With reporting by Rachel Zelkowitz.