ResAIKit
Research Integrity Toolkit
Back to the encyclopedia
W3Image forensicsWestern BlotLayer 1 (Deterministic)

Background Tiles

Detects backgrounds that were built rather than captured. A real western blot background carries film or sensor noise and never repeats; a background that is artificially flat (lacking that noise) or that repeats a tile pattern from copy-paste editing is fabricated. The indicator masks the bands, measures the local noise of the background, and looks for periodic repetition in its autocorrelation.

Technical description

Detects and masks the bands, tiles the remaining background into a 16x16 grid, and reads two properties. Local noise: the median within-block standard deviation; below 1.5 on the 0-255 scale flags an artificially flat fill, since a real background carries a noise floor even when its mean is constant. This replaces the coefficient of variation of the block means, which is also small for a genuinely noisy background and would misflag it. Periodicity: the normalized autocorrelation of a central crop, with the centre zeroed; three or more points above half the peak indicate a tiled pattern. Each signal adds up to 2.5, capped at 5.0.

How it works

Layer 1 (deterministic): masks detected bands, then flags an artificially flat background when the median within-block noise falls below 1.5, and a tiled background when the autocorrelation of a central crop shows three or more peaks above half its maximum. Each signal adds up to 2.5 and the score is capped at 5.0; findings describe a flat or periodic background.

Why this matters

The background of a blot records the physics of exposure: a captured background carries a noise floor from the film or sensor, and the absence of that noise is a clear sign that a region was painted in, the same noise-inconsistency principle behind blind image forensics. Repetition is the other tell: copy-pasting a background patch to cover a deleted band or extend a field leaves a periodic signature the eye misses but autocorrelation reveals. Erasing or constructing background content is recognised misconduct, and detection must look beyond the bands to the spaces between them.

Score thresholds

0-1
The background carries natural noise and shows no repetition
2-3
The background is artificially flat, or shows some periodic structure
4-5
The background is both flat and periodic, or strongly tiled, consistent with a constructed background

Limitations

The screen depends on detecting and masking the bands, so missed bands leave band pixels in the background and oversized masks remove real background. The local-noise test assumes a real background is noisy, so a heavily denoised or strongly compressed image can read as artificially flat without being painted. The periodicity measure counts points above a threshold rather than isolating discrete peaks, so a smoothly varying background with strong uneven illumination has a broad central correlation lobe that can suggest periodicity that is not there. The wrap boundary can create edge artefacts. Duplicated bands are indicator W1 and sharp transitions around individual bands are indicator W4; W3 stays on the flatness and repetition of the background.

References

  1. Mahdian B, Saic S. (2009). Using noise inconsistencies for blind image forensics. Image and Vision Computing 27(10):1497-1503
  2. Bik EM, Casadevall A, Fang FC. (2016). The Prevalence of Inappropriate Image Duplication in Biomedical Research Publications. mBio 7(3):e00809-16
  3. Cromey DW. (2010). Avoiding twisted pixels: ethical guidelines for the appropriate use and manipulation of scientific digital images. Science and Engineering Ethics 16(4):639-667