ResAIKit
Research Integrity Toolkit
Back to the encyclopedia
M8Image forensicsMicroscopyLayer 1 (Deterministic)

Charging Artifacts

Reads the saturated bright and dark regions of a scanning electron micrograph and asks whether they look like genuine charging. A non-conductive specimen produces irregular, organic saturated zones; geometric shapes such as filled circles are the mark of a drawn cover, and a total absence or an excess of saturation is also recorded.

Technical description

Counts saturated pixels (above 250 or below 5); a total absence adds 2.0. When saturation is present, the saturated regions are extracted as contours and their shape regularity is measured with the isoperimetric circularity 4*pi*A/P^2 (one for a circle, lower for irregular outlines). The contour circularities are combined into an AREA-WEIGHTED mean, so a single large drawn region stands out among small organic spots rather than being averaged away. An area-weighted circularity above 0.7 adds 2.5 (geometric), below 0.4 adds 0 (organic), and between adds 1.0; a saturated ratio above 0.1 adds 1.0. Contributions are capped at 5.0.

How it works

Layer 1 (deterministic): counts saturated pixels, and if present extracts their contours and measures circularity, combining contours into an area-weighted mean so large drawn regions dominate. Geometric shapes (area-weighted circularity > 0.7) add 2.5, organic shapes add 0, and ambiguous ones add 1.0; a total absence of saturation adds 2.0 and excess saturation above 10 percent adds 1.0. The contributions are summed and capped at 5.0.

Why this matters

Charging is a defining, hard-to-fake feature of scanning electron microscopy on non-conductive specimens: accumulated charge deflects the beam and modulates emission, producing irregular bright and dark artefacts whose boundaries follow surface conductivity and topography, not a drawing tool. A saturated region with a smooth circular or rectangular outline is therefore out of place and points to a drawn cover hiding content, a manipulation that is misconduct even when locally seamless.

Score thresholds

0-1
Saturated regions are present and organic, consistent with genuine charging
2-3
One anomaly: no saturation at all, an excess of saturation, or ambiguous saturated shapes
4-5
Geometric saturated shapes, alone or with an absence or excess of saturation, consistent with a drawn cover or an edited region

Limitations

The absence-of-charging cue is the weakest, because charging is not universal: a conductive specimen, a sputter-coated sample, a low-vacuum or environmental SEM, or a well-tuned acquisition can legitimately show little saturation, so a clean image is flagged only weakly and should prompt review rather than be taken as proof. The shape cue assumes a manipulator draws a regular outline, so an irregular cover evades it and a compact genuine spot can raise the circularity. Digital-contour circularity is biased by boundary staircasing, mitigated but not removed by the area weighting. The saturation thresholds are fixed. The spatial noise profile is handled by indicator M6, so M8 stays on the saturation regions and their shape.

References

  1. Cazaux J. (2005). Recent developments and new strategies in scanning electron microscopy. Journal of Microscopy 217(1):16-35
  2. Cox EP. (1927). A method of assigning numerical and percentage values to the degree of roundness of sand grains. Journal of Paleontology 1(3):179-183
  3. Bik EM, Casadevall A, Fang FC. (2016). The Prevalence of Inappropriate Image Duplication in Biomedical Research Publications. mBio 7(3):e00809-16
  4. Rossner M, Yamada KM. (2004). What's in a picture? The temptation of image manipulation. The Journal of Cell Biology 166(1):11-15