As Horizon 2020, the European Commission’s 7-year research and innovation framework, enters its final year, attention turns to Horizon Europe, the follow-on program that will run 2021 to 2027. The $111 billion research and innovation framework program focuses on open science – public visibility and open access to scientific publications and research data.
The Open Microscopy Environment (OME) consortium, a consortium of academics and industry experts that creates open-source software and standards, has already established standards in microscopy for open science, aimed at the use of findable, accessible, interoperable, and reusable data. Those standards could deliver cost savings, greatly accelerate research and innovation, improve research quality, and provide researchers in Europe with a substantial competitive lead. At Leica Microsystems, the transition to open collaboration is already occurring.
Standard formatsIn the future, Leica scientists could extend the basis of their own image data by including pictures of the same sample from third parties to help drive progress. For example, researchers could perform evaluations on a broader statistical basis in less time or concentrate entirely on the evaluation of existing images based on new points of view.
Putting aside differences in sample preparation, a standard image file format in a quasi-common language for image information would be helpful for execution.
Previous attempts by various microscopy manufacturers to develop software to collect or convert high-volume imaging data to OME-compliant metadata were hampered by the variety of parameters used by manufacturers and their operating software for the data’s collection, documentation, and accessibility.
Comparability and reproducibility are major challenges. To harmonize designations, manufacturers need to collaborate to ensure the same parameters are managed in the same place with the same scaling. Proprietary imaging functions remain difficult to use if results are directly contained in the image information. Developers must balance progressive innovation in image acquisition with the desire for standardized evaluation across device classes. Even though the OME format administers device-specific parameters, image evaluation across device classes remains difficult as specific parameters are only applicable for a share of the image data. Parameters describing the characteristics of fluorescence hardware filters, for example, are different from parameters of freely tunable spectral detectors.
Manufacturers and researchers must determine what they want to achieve with format standardization because they cannot cover the full range of data. Manufacturers constantly extend file formats supporting the latest advancements of an instrument. However, Leica Microsystems and other manufacturers could support original OME formats by allowing scientists to save and load to one standard format. Users must find the point of the best possible conversion from original file format to a standardized OME-compliant format because, right now, there is no perfect match.
Since format consolidation talks began, technologies such as machine-learning have advanced, making it clear that scientists would embrace OME-compliant metadata collection or conversion industry-wide. Precedents show metadata’s power and utility, particularly when combined with machine learning or artificial intelligence (AI) algorithms. Machine learning using digital fundus photography and optical coherence tomography, for example, can automate early detection of eye diseases such as retinopathy and macular edema in people with diabetes. Powerful graphics chips (developed for video gaming) have advanced parallel processing power, boosting the use of AI in medical imaging by supporting computationally intensive systems known as artificial neural networks. Combining artificial neural networks, processing power, and massive image data sets allows researchers to create deep-learning networks that can perform more sophisticated tasks.
With these advances progressing at an exponential rate, the time seems ripe for a renaissance of projects working toward OME-compliant metadata collection and/or conversion. The question is no longer if, but when and how.
While many companies believe collaborative, open innovation is key to success, others continue to think that keeping information private leads to proprietary privilege.
In a sector as competitive and sensitive as healthcare and medical science, this can be difficult. The key is finding trustworthy partners who can openly collaborate to identify and define a problem and discover and develop the right solution.
Speaking the same language
A key hurdle to industry-wide OME-compliant compatibility is the metadata file format. Differences and compatibility issues between raw imaging data file formats should be addressed, or the vast array of file formats used in microscopy and imaging could hinder open collaboration. For instance, all manufacturers can save data in an OME minimal accessible file format, but the information is not directly stored that way, requiring additional actions to convert files. Important information is often removed due to the minimal definition.
OME developed three collaborative tools; Bio-Formats, OMERO, and OME Files. However, when the software was tested by manufacturers and laboratories, manufacturers and users could not come to a consensus regarding a single cohesive format and operating language that minimized changes to the original file format. This, however, has not deterred the OME consortium of universities, research labs, industry, and developers in their efforts to produce open-source software and format standards for microscopy data. If anything, the willingness of the testers highlights the need and market for these tools. Moreover, given the current environment as we embark on the fourth industrial revolution, it is more pressing and pertinent to rejoin the effort to develop and implement platforms for collecting, converting, and openly sharing OME-compliant metadata.
One solution could be proactively sharing learnings and insights on open platforms in any language to increase collaboration, eliminating the possibility of valuable information being removed in the conversion. Earlier this year, the Public Library of Science (PLOS) Open Biomaterials Research collection launched an open platform for the biomaterials community to submit research under the premise of open science. Research submissions should include relevant protocols and recipes, fully accessible data and source codes, and any other practical details critical to the success of the experiment or project. An editorial board then oversees submissions to ensure the platform remains a meaningful and collaborative resource for the community.
That sort of collaboration may be better geared toward researchers than microscopy manufacturers. How then can industry actors promote open science and innovation? One approach is to partner with third parties. A recent survey by the Boston Consulting Group (BCG) found that company partnerships with startups rose from 59% in 2015 to 75% in 2018. Similarly, partnerships with academia increased from 60% to 81%, and partnerships with fellow companies grew from 65% to 83%.
Collectively, we can identify the priorities of microscopy users when it comes to imaging data and metadata, factoring in current and future needs and preferences. We can discover common needs and objectives and arrive at an overlapping consensus for metadata file formats and the practical particularities of open sharing platforms and rights and protections of all involved in the provision, sharing, and use of data. It is fundamental for as many stakeholders as possible to participate, find the solutions together, and ensure that the future we inform and shape together is one that we are all satisfied with.