An award-winning microscopy video created by an optical engineer at the National University of Singapore has ignited a scientific controversy after researchers questioned whether its striking visual elements were generated by artificial intelligence. Nikon Instruments later updated its contest blog to state that an unsupervised AI model was used during post-processing.
A prestigious microscopic videography competition has found itself at the center of a debate over digital manipulation and transparency in scientific imaging. The annual contest, run by Nikon Instruments, celebrates footage captured through light microscopes to showcase the hidden complexity of biological systems.
On September 15, optical engineer Ning Xu was named the winner of the competition for a video depicting the movement of hair-like structures known as cilia from a child with a rare genetic disorder.
Microscopy Experts Question the Biological Accuracy of Contest Winner Ning Xu
Days after the announcement, researchers took to social media to scrutinize the award-winning clip. Edward Phelps, a bioengineering researcher at the University of Florida in Gainesville, pointed out anomalies in the video on LinkedIn.
The purple structures resemble mitochondria but such a sub-epithelial structure composed of extracellular mitochondria of the same size as cell nuclei does not occur in biology.
Edward Phelps, bioengineering researcher at the University of Florida in Gainesville, via Nature
Phelps noted that while blue structures in the video resemble cell nuclei, they do not behave like them, and the purpose of the red stain remains unclear.
Melanie White, a developmental biologist at the University of Queensland in Brisbane, Australia, emphasized the critical need for fidelity in scientific visuals. Scientific images are not just illustrations, she noted.
They are data, and we need to be able to trust that what we are seeing is grounded in the underlying measurement.
Melanie White, developmental biologist at the University of Queensland in Brisbane, Australia, via Nature
Markus Sauer, who studies super-resolution microscopy at the University of Würzburg in Germany, added that utilizing AI tools to visualize experimental data is not inherently problematic. However, he cautioned that difficulties arise when those tools alter, exaggerate, or misrepresent the underlying findings.
Contest Organizers Update Blog as Creator Defends Visual Enhancement Choices
Following the online discussion, Nikon Instruments updated its contest blog on September 22 to disclose that an unsupervised AI model was used to assist with post-processing on the winning entry.
Xu addressed the criticism on LinkedIn, explaining that his team applied digital enhancement only after reconstructing the grayscale data.
Ning Xu explained that AI was applied afterwards to the reconstructed grayscale data to distinguish and color structures with similar morphology, adding that because the video was prepared for a competition celebrating the beauty of microscopy, the team wanted the final presentation to be visually engaging as well as scientifically interesting.
Xu stated that artificial intelligence was not employed to generate the experimental movie, the cilia themselves, or their motion. He noted that the post-processing was conducted without making anatomical claims about what those rendered features represent.
Broader Scrutiny Surrounds Digital Manipulation in Scientific Imaging Contests
The controversy underscores increasing vigilance regarding digital manipulation across academic journals and public showcases. As imaging technology and sophisticated editing tools continue to advance, competitions relying on participant disclosures and the honor system face mounting pressure to evaluate their verification standards.
White observed that it remains unclear how faithfully the submitted video represented the actual microscopic data, raising the possibility that the AI tool introduced biological structures absent from the original measurements.