Last week, Nikon Instruments Inc. declared the winners of the annual Nikon Small World in Motion contest, which celebrates films captured through microscopes. Ning Xu, an optical engineer at Tsinghua University, bagged first place for a video showing abnormal beating of cilia in the airway of a child with primary ciliary dyskinesia (PCD), a rare genetic disorder that affects respiratory function.
However, soon after this announcement, several researchers alleged the use of AI in creating some parts of the video, raising concerns about the undisclosed use of AI.
For Robert Hirst, a lead scientist who uses microscopy to diagnose PCD at Leicester University, the film immediately set off alarm bells. “When you start looking at the scales of the movie and some of the details of it, it was quite clear to the community…that some of these features just are not consistent with nature,” he said.
In an email to The Scientist, a Nikon Instruments Inc. spokesperson said that the organizers have reviewed supporting materials provided by Xu. “Based on the information currently available, Nikon can confirm that an AI-assisted enhancement tool was used by the entrant as part of post-processing. Nikon determines this particular use of AI—as a tool for enhancement of microscopy-generated source data, not for image generation from scratch—to be in compliance with the rules set forth by the competition,” they said. The organizers also updated the winning entry page to declare that AI was only used in post-processing of the image, a stand that Xu maintains.
Winning Video Comes Under the Microscope
For the video in question, Xu subjected a sample derived from a patient with PCD to a novel technique called diffractive super-resolution microscopy through a custom optical microscope, which yielded data in greyscale. “Then, we used an unsupervised neural network for post-processing and visualization,” explained Xu. “It [helped] us bring out features in the grayscale data and make the final video more visually engaging.”
The Nikon Instruments, Inc. spokesperson said, “The entrant has provided detailed technical documentation outlining the microscopy equipment, imaging methods, and processing techniques used to create the source video. Nikon’s team found the described methods technically legitimate and consistent with established microscopy practices.”
However, Christophe Leterrier, a neuroscientist at the French National Centre for Scientific Research-Aix Marseille University, was not entirely convinced. “The processing has been done in a way that has altered the original image too much,” said Leterrier, former Nikon Small World jury member and multiple-year awardee.
According to Nikon Instruments Inc., the video showcases a single-cell section at extreme magnification. But Leterrier said that the layer beneath the cilia shows what specialists have identified as a number of small nuclei, mitochondria, and plasma membrane. “You have something very nice and very engaging visually but then it’s absolutely not representing a biological reality,” he said.
Researchers Flag Biological Inconsistencies
Since Hirst first noticed the discrepancies, he got in touch with Nikon Instruments Inc. and Xu over social media. Despite their back-and-forth in the comments, Hirst is convinced that AI has been used in generating some aspects of the video. “The big key here is [they have used an] unsupervised neural network,” he said. According to him, the film misleads viewers into thinking that there is one cilium per cell, when in reality, there are about 200. “It’s created this substratum underneath the cilia which is wrong in scale, wrong in biology, and just looks completely fabricated.”
Melanie White, a cell biologist who uses quantitative imaging at the University of Queensland, shared Leterrier’s and Hirst’s concerns. “The image itself, although very beautiful, has raised questions about the biological observation that it proposes to display,” she said. “It’s very difficult to understand if what is being shown actually…faithfully represents the biology.”
Xu maintained that he used AI only to process and not generate any parts of the movie. “There is an important difference between using AI to create something that was never in the experiment and using AI to process or visualize real experimental data,” he said. “The important question is not simply [if] AI was used; we should ask how was it used.”
Despite the controversy, Xu found researchers’ concerns “understandable.” Xu added, “AI is developing very fast and people want to know what is real and what has been changed by AI.”
Zooming Out: The Big Picture of AI in Microscopy
To Hirst, the fact that this video represents a disease is troubling. In people with PCD, cilia movement is often altered, preventing the airways from getting cleared and increasing the risk of recurrent respiratory infections. Researchers rely on deciphering the movement of cilia, not just their appearance, for diagnosis. “What worries me most of all is that PCD patients will see this video and think, ‘My cilia look like this, and my cells look like this,’ and they just don’t, unfortunately.”
Leterrier said that the community’s problem is not with AI, but with the undisclosed use of AI to this extent. “AI is…tremendously useful to our work as microscopists,” he said, but cautioned that the boundaries of AI use and disclosure of the extent to which AI tools were used have to be clear.
White agreed that there are useful applications of AI tools in microscopy. “[But] there is a difference between using AI to enhance information that was already present or captured versus using AI to interpret and create new information that was not initially there,” she said. “It’s difficult to understand where Nikon is drawing the line around the use of AI.”
Amidst increasing pushback from the community, Nikon Instruments Inc. updated the movie caption, calling it a rendition. In response to whether their stance has changed, the spokesperson said, “We are reviewing the AI policy. At this time, we don’t see that any rules were violated, but we are reinvestigating and re-evaluating the entry.”
Leave a Reply