Mapping bioimage-analysis software adoption in microscopy
This study analyzes 1.7 million microscopy papers from 2010 to 2025 to reveal a growing reliance on image-analysis software—led by ImageJ/Fiji with increasing field-specific adoption of tools like QuPath and napari—while highlighting that citation counts often underestimate actual usage and urging researchers to explicitly cite and report their software methods to improve reproducibility.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Modern microscopes do not just take pictures; they generate vast, multi-layered digital files that hold the secrets of how cells move, divide, and interact. However, a raw image is only the beginning of the story. To turn those pixels into biological knowledge—counting how many cells are present, measuring their size, or tracking their movement—scientists must use computer programs to analyze the data. These programs are the tools that translate visual information into numbers. The choice of which tool to use matters deeply because different programs can sometimes produce different measurements from the exact same image. If two laboratories study the same biological event but use different software, they might reach different conclusions about the size of a cell or the number of times it divided. This creates a challenge for science: if the tools are not clearly named and described, other researchers cannot repeat the experiment to verify the results.
For years, the scientific community has known that these software tools are critical, yet no one had a clear map of which ones were actually being used in the real world. The information was scattered, hidden inside the methods sections of millions of research papers, often mentioned in passing without a formal reference. To solve this, a team of researchers set out to count and categorize the software mentioned in the global literature of microscopy. They scanned 1.7 million full-text research papers published between 2010 and 2025, looking for mentions of thirty specific software packages. Their goal was to see how the landscape of these tools has changed over time, which ones have become the standard, and how researchers are reporting their use.
The researchers found that the use of image-analysis software has grown much faster than the number of microscopy papers themselves. In 2010, only 12 percent of microscopy papers mentioned at least one of the analysis tools they were tracking. By 2025, that number had risen to 30 percent. Despite this explosion in usage, the field has not become a chaotic mix of hundreds of different tools. Instead, the mentions remain heavily concentrated on a few general-purpose platforms. One program, known as ImageJ/Fiji, dominates the landscape. In 2025, it was mentioned in 28 percent of all full-text microscopy papers, accounting for 82 percent of all mentions of the analysis packages they tracked. This suggests that while new tools are appearing, the scientific community still relies on a small set of established, versatile platforms for the majority of its routine work.
However, the study also revealed that the dominance of these older platforms is slowly shifting as specialized tools gain ground in specific fields. For example, a program called QuPath has seen rapid growth, particularly in the field of pathology, where scientists study tissue samples to diagnose disease. Similarly, a tool called napari is frequently mentioned in papers dealing with electron microscopy, which uses beams of electrons to see structures far smaller than light can reveal. Newer methods based on deep learning, which allow computers to learn how to recognize patterns like cell boundaries, are also becoming common. Yet, even these advanced methods are often used alongside the established platforms. The researchers found that papers mentioning the deep-learning tool Cellpose often also mention ImageJ/Fiji, and papers using StarDist frequently run it inside the ImageJ environment. This indicates that scientists are integrating new, powerful algorithms into the familiar workflows they already know, rather than abandoning their old tools entirely.
A significant portion of the study focused on how well researchers are reporting their methods, a key factor in scientific reliability. The team discovered that many papers mention a software package without actually citing the paper that introduced it. For the most popular tool, ImageJ/Fiji, 86 percent of the papers that mentioned it in 2025 did not include a citation to its original publication. This gap is even wider for newer tools like QuPath and Cellpose, where the share of papers mentioning the software without citing it has risen sharply. This means that simply counting how many times a software is cited in a reference list underestimates how many people are actually using it. Furthermore, the researchers audited the sentences where software was mentioned and found that a large number of papers failed to provide essential details. In 2025, 61 percent of papers mentioning ImageJ/Fiji, 52 percent of those mentioning StarDist, and 36 percent of those mentioning Cellpose did not specify the version of the software, the specific function used, or the settings applied. Without these details, it is impossible for another scientist to know exactly how the analysis was performed, making it difficult to reproduce the results.
The study concludes that while the number of papers using image-analysis software is growing, the field remains anchored by a few core platforms that researchers trust for their versatility. The rise of deep learning has added powerful new capabilities, but these are often woven into existing workflows rather than replacing them. The most pressing issue identified is not the lack of tools, but the lack of transparency in how they are used. The authors argue that for science to remain reproducible, researchers must do more than just name the software they used; they must cite the original papers, specify the version numbers, and describe the settings and parameters. By making these details visible, the scientific community can ensure that the digital tools driving modern microscopy are as reliable and transparent as the microscopes themselves.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.