Auditing figures for salami slicing via reverse image search

Methodology

The text of a paper is almost always polished to a mirror finish. The visual data, however, is where the seams usually show. I remember the 2016 fallout from those duplicated western blots in the oncology journals; the authors claimed they were the same biological sample, but the background noise was an exact pixel match. It was a textbook case of lazy recycling. If you suspect a paper is just a slice of a larger study fragmented to inflate a CV, stop reading the abstract. Take the control blots or the primary figures and run them through a reverse image search. Google Lens is sufficient for blatant repetitions, but tools like ImageTwin are better for spotting rotated, cropped, or mirrored duplicates. The process is simple: isolate the control image, upload it, and check if that same data appears in an unrelated publication. It moves the audit away from the carefully curated narrative and toward the raw evidence, which is much harder to rewrite.

4 comments

Comments

ThreadDiggerTess·1 hour ago

This is significantly more effective when paired with mandated raw data deposits in repositories like Figshare. Searching original TIFF files avoids the compression artifacts that often mask clones in final PDFs.

LurkingLorraine·1 hour ago

google lens misses mirror-flipped blots and often fails on grayscale noise.

DevilsAdvocate_Dan·1 hour ago

What if the author is publishing a linked series of papers? In that case, using the same control blot might be a requirement for internal normalization across the set.

ProfActuallyPhD·1 hour ago

The stochastic nature of chemiluminescence means the background grain acts as a unique fingerprint. When you see identical pixel-level noise in separate figures, the probability of that being a biological coincidence is effectively zero.