key: cord-0611714-o6xypqik authors: Hohlbein, Johannes; Diederich, Benedict; Marsikova, Barbora; Reynaud, Emmanuel G.; Holden, Seamus; Jahr, Wiebke; Haase, Robert; Prakash, Kirti title: Open microscopy in the life sciences: Quo Vadis? date: 2021-10-26 journal: nan DOI: nan sha: d8bf99c003902ac32c8955adb5d0b7c0c7c851bd doc_id: 611714 cord_uid: o6xypqik Light microscopy allows observing cellular features and objects with sub-micrometer resolution. As such, light microscopy has been playing a fundamental role in the life sciences for more than a hundred years. Fueled by the availability of mass-produced electronics and hardware, publicly shared documentation and building instructions, open-source software, wide access to rapid prototyping and 3D printing, and the enthusiasm of contributors and users involved, the concept of open microscopy has been gaining incredible momentum, bringing new sophisticated tools to an expanding user base. Here, we will first discuss the ideas behind open science and open microscopy before highlighting recent projects and developments in open microscopy. We argue that the availability of well-designed open hardware and software solutions targeting broad user groups or even non-experts, will increasingly be relevant to cope with the increasing complexity of cutting-edge imaging technologies. We will then extensively discuss the current and future challenges of open microscopy. Light microscopy has been pivotal in the life sciences to study small features and objects otherwise hidden to the naked eye. Simple microscopes such as the Foldscope or smartphone auxiliary lenses are available for a few Euros and form the basis of citizen science projects, scientific education, and medical diagnosis 1,2 . technology, more and more information on modalities for advanced and high-end microscopy has become openly available to interested researchers and private enthusiasts. Compared to the open software movement, however, the development of open hardware frameworks for microscopy accelerated only rather recently with the increasing accessibility and affordability of suitable components. With 3D printers, rapid reproduction of designs and prototyping moved from professional machine shops to the living room table. Web shops deliver sophisticated CNC milled designs within days to the doorsteps. Mass-produced electronics such as light-emitting or laser diodes, microcontrollers, optical parts, and (mobile phone) cameras have significantly levelled the playing field for high-end instrumentation. Open microscopy does not automatically imply that solutions are cheap. Some of the frameworks further discussed below feature higher-priced components such as microscope objectives, laser engines, and cameras often costing tens of thousands of Euros. As such, these components maximize performance but can surprisingly often be adequately replaced by cheaper alternatives that better benefit from the economies of scale 9 . kept the science largely behind closed doors; although results were published, information on the methodology, the experimental settings or access to raw data was, and unfortunately often still is, hard to come by ( Figure 1 ). Open science is addressing these issues by providing a myriad of additional interaction points between researchers and more broadly between citizens, starting from the collection of ideas over reusing hardware designs and open-source software to the exploration of publicly available data by anyone interested. For scientific data, the FAIR principle (Findable, Accessible, Interoperable, Reusable) 11 provides guidelines for moving towards what science should be all about: shared knowledge accessible to all. With that in mind, we here define open microscopy as scientific data which contains information on (i) how to build, maintain and use (light) microscopes, (ii) how to prepare, handle and measure samples and (iii) how to analyze, store and distribute experimental data and models. For open hardware, recent work highlights general opportunities and best practices 12, 13 . For light microscopy, this implies making new imaging technologies accessible and easily reproducible. In the past, new methods were often published with only a vague cartoon of the optical setup and without access to any software or electrical circuit diagrams. Even for specialist labs, implementation of published modalities often required extensive reverse engineering and tinkering for many years. Commercial microscopes, on the other hand, are often built on closed-source methods and compound the reproducibility problem by making key steps in the process of biological discovery dependent on an ill-characterized black box. Open microscopy can address this problem by ensuring that any new imaging method, both hardware and software, is sufficiently documented and open to allow straightforward reproduction by other specialists, and without information being hidden behind restrictive Material Transfer Agreements (MTA) or Non-Disclosure Agreements (NDA). Documenting assembly and manufacturing (e.g., 3D printing, CNC machining) guides, bill of materials or even providing video tutorials that help others recreating a setup without additional help by the authors, acts as an amplifier, in which knowledge is multiplied and can be used at remote places outside the originating lab. It can be very motivating to see one's work being replicated and used by others, even if current scientific scoring systems may only recognize formal citations and not pre-publication replications. Since prototype microscopes usually require extensive further development to make them production-ready and user-friendly, this does not preclude their co-existence with commercial devices. The need for users to receive comprehensive product support and a turnkey solution is not necessarily a requirement of the Open Microscopy movement. Generally, we argue that open microscopy should be considered an essential requirement for any scientific publication of a new microscopy method to meet modern standards of scientific reproducibility. Detailed documentation, including why something was implemented in the suggested way, is also key for others to learn about the given technique and to discover potential optimization steps. For small hardware or software components, this could further imply making drawings or source code available under an open-source license. In general, documentation can be written in the form of citable scientific publications and thus, is well compatible with the academic incentive system. Editors from the optics and scientific imaging fields, however, should increase efforts in supporting scientists, who are willing to share their designs openly, to improve the reproducibility of publications and science in general. Reproducibility in the context of data analysis implies that both the experimental or simulated data as well as the analysis software is openly available. We therefore argue that reproducibility is the key to advancing science as without rigorous verification of results and discoveries true scientific progress is impossible. Academic researchers should be aware that, by default, everything developed and Until recently, microscopy hardware developers seeking to develop new optical methods faced the choice of either retrofitting new hardware onto an existing commercial microscope (body) or designing and building an entire bespoke microscope from individual optical components. Neither fits the requirements of a hardware developer perfectly. Although using a commercial body has the advantage of starting with a working system that usually lacks serious optical aberrations and offers a stable mechanical base, critical optical planes required in the microscope development process are often hidden, unknown or inaccessible within the monolithic body. Furthermore, some features implemented for user-friendliness and safety (eyepieces, safety interlocks, dedicated software) can potentially prevent the hardware developer from easily modifying a setup. Customized designs, on the other hand, offer full control and often lower costs at the expense of significant development time and the need to re-implement basic components and features, such as focusing or sample positioning. One drawback of custom microscopy solutions is that they are often less user-friendly compared to their commercial counterparts that offer safe housing or one-click software solutions. In the end, commercial bodies and self-designed systems may target different audiences but, as we will discuss in the following, the gap has been closing rapidly. Rather than classifying open hardware frameworks in terms of their apparent complexity, we suggest looking at potential user groups. We foresee disruptive potential of Although involved developers have a broad set of goals and applications, a common conviction can be found: the principles of open science have the power to change microscopy for the better, leading to more innovation, accessibility, and reproducible science. An implementation of open microscopy in its purest form in which ideas and experimental designs are shared online and on social media sometimes even before experimental realizations take place has been pioneered by Andrew York (https://andrewgyork.github.io/). In the following, we discuss some of the key challenges. The One often-used statement in the open microscopy field is that an instrument was designed for (part-) costs that are ten times cheaper than the price of a comparable commercially available instrument. We consider such statements at best misleading and at worst plainly wrong as neither the costs of development nor the time spent to build the instrument is accounted for in most cases. We also point out that any company must fulfil a minimum of conformity with health, safety, and environmental protection standards (e.g., CE, FCC, TÜV or others) for their products and provide customer support. In open projects, even when building up with commercially available components, the sole responsibility on safety is shifted to the user. Additionally, customer support depends largely on the goodwill of the developer. As outlined above, the number of open microscopy frameworks has hugely increased over the last years. An interesting example of proliferation is the variety of software packages available for single-molecule localization microscopy that fit data from individual fluorescent emitters to obtain their position with sub-pixel accuracy. Recently, over 30 different software packages were compared using a diverse set of metrics 43 . Although such wide choice can be intimidating at first, the robust assessment itself, indicating strengths and weaknesses of individual software packages, is excellent proof of the benefits of open microscopy. As new packages and developments can be directly compared to previous ones, the super-resolution field was able to quickly reach a high degree of maturity in terms of data analysis. It became equally apparent that user-friendly implementations of the best approaches are critical to reaching end-users. 45, 46 . We suggest that these best practices are requested and followed by scientists, reviewers, and editors to ensure and enable long-lasting inter-device operability. We note, however, that hardware-and software development significantly differ in the way how solutions are distributed. Ready-toinstall software packages require only a limited knowledge about the underlying system integrity. Well-formulated building instructions still require someone who gathers all necessary parts (e.g., 3D printed or from external sources), assembles these parts and invests a fair amount of time to finally get the system running -no matter how well the documentation is described. The equivalent to a ready-to-install software package could be assorted hardware kits (UC2 4 ), but the non-zero cost of duplicating hardware poses a challenge for widespread distribution. It remains important to implement and use calibration control points and sample standards both for hardware as well as software. Open microscopy projects are shared using different files and data formats. Whereas many formats are open and suitable viewers are freely available, this is not necessarily the case for designs that feature Computer Aided Design (CAD). For CAD, buying suitable software licenses can be prohibitively expensive leading to limited accessibility and unknown design history. For 3D printing, 3D models can be exported, for example, in the *.stl format which describes only the surface geometry of a three-dimensional object without any scale thereby inhibiting any modifications on the design. Alternatives, such as sharing links to cloud-based CAD software (e.g., Fusion360, Tinkercad) or relying on open-source CAD models (e.g., openSCAD, Python-based FreeCAD) can help to distribute design files across many different development environments. Ultimately, publishers and developers should ensure that design files are available as (vectorized) raw data-file as well as processed files. The open source hardware association (OSHWA) offers a guideline to maintain a good quality of openness in projects (https://www.oshwa.org/sharing-best-practices/). Whilst protein structures, DNA sequences and other bioinformatic data can be deposited in well-established databases (e.g., PDB, EBI), no dedicated repositories for microscopy hardware are currently available. The close connection between open hardware and software is inevitable for complex microscopy projects. Projects such as µManager 35 , Pycro-Manager 36 and Python-microscopy 37, 47 have been playing a key part in connecting setup control, data acquisition and data analysis. When it comes to hardware control, the availability of open-source device drivers and adapters is crucial. The software architecture used in µManager, for example, standardizes how hardware devices can be controlled from diverse software components via a plug-in mechanism. If this mechanism is well-designed and easy to use, developers from all over the world will contribute plugins. As a case in point, the µManager community managed to collect hundreds of device adapters (https://micro-manager.org/Device_Support). Many of the earlier introduced approaches for computational microscopy have the potential to lower the quality requirements for microscopy hardware. Efficient denoising algorithms, for example, allow the use of lower laser powers for fluorescence excitation or the use of cheaper cameras without compromising final image quality. reliably into any open microscope control software yet. Combining open software solutions for microscope control, image processing and data analysis is hugely challenging, requiring many developers from different backgrounds closely working together to optimize signal and data streams. First promising steps have been made by the smart microscopy software autopilot 55 and by combining OpenFlexure, ImJoy and UC2 56 . These scenarios highlight how important interdisciplinary collaboration is when it comes to solving problems in biology using experimental imaging and software approaches. Overall, plugin-based software projects that provide a reusable development platform can help integrating, e.g., graphical user interfaces (see Fiji, napari, ImJoy). Plugins can be easily shared with the community; algorithms and code can thus be used without much prior knowledge, which leads to a significantly increased acceptance by users. Open-source licenses enable the free (re-)use of software and thus contribute to accelerated research, as solutions are not always created from scratch. From our experience, many microscopy projects are initially driven by one or two people and have a limited time as scientific advancements can render ideas and concepts obsolete very quickly. Some projects, however, develop into large community-driven projects with specific requirements to gain and maintain long-lasting relevance and impact. When it comes to questions regarding long-term support, clear communication between the developer and the potential target audience is advised to keep expectations aligned and in check. Developers should indicate as soon as possible whether their project is intended as a research platform for others that could turn into a community-driven project or whether the developer is mainly interested in using their hardware or software to promote their research. Furthermore, interacting with the community, selecting issues to work on and motivating others to support an open hardware initiative requires a substantial investment of time and effort. We recommend developers who want to develop community standards to think about these aspects carefully and identify sources and people for support early, noting that follow-up costs, both in time and money, cannot be paid by a single PhD student or postdoc no matter how enthusiastic they are. What are the requirements for microscopy projects to succeed as community standards? First and foremost, we think that a clear need should be identified by the developer/community. As a recent example, the need for an adaptable image viewer interacting with the Python environment kickstarted the impressive and community-driven development of napari (https://zenodo.org/record/3555620). Second, some uniqueness to the approach differentiating it sufficiently from existing solutions should exist. If uniqueness is not sufficiently given, we recommend contributing to the other project. Third, one or more core developers with sufficient resources in terms of time, money, or appreciation are required to ensure continuity. Fourth, create and maintain an active user based on all levels of involvement ranging from "use as is", "test and report bugs", "request features" to "fix bugs and implement small features". The "miniscope"-community offers a great example how the base assembly, a miniaturized fluorescence microscope that enables in vivo brain imaging of mice, is used by the community to add a variety of soft-and hardware addons, such as two-photon imaging of oneshot 3D imaging using lightfields further reviewed in reference 57 . We note, however, that "open source" should not be translated as "free support". People already involved in a project have to talk to other potential users, help them, create examples, code/develop for them, work in close contact with them, find out what they want, create a product what they are willing to use without extra effort, give them the feeling of "YOU DID IT!". Fifth, merging expertise by means of adapting hardware or software designs from different projects, such as taking advantage of the microscopy OpenFlexure stage within the UC2 system can speed up development processes. Sixth, developers should strive for device interoperability by means of openly developed interfaces. Here, the software community gives a nice example of how objectoriented and packaged code enables simple reuse and hence speeds up development. Seventh, extensive high-quality documentation should be seen as a core priority. This documentation allows new users and developers to easily join and continue a project even if initial contributors left or initial investments have run dry. We note that larger imaging facilities are well suited to support developers and users. We hope that universities and funders recognize the potential value of having a wide portfolio of maintained open microscopy projects. We think that it can be advantageous especially for hardware projects if (parts of the) assemblies could be made commercially available. We see an increasing demand for affordable and proven high-quality microscopy solutions by end-users who are not interested in building scientific instrumentation. In the simplest case, 3D printed or CNC-milled entities (e.g., OpenFlexure or miCube) can be sold directly, or prototypes can be made available that still require further adjustments to function to the user's need in the form of do-it-yourself kits. In other cases, microscopy solutions could become refined and user-ready products. For this route, however, there are many challenges to overcome. First, investors required to finance the transition from a prototype to a full product generally prefer solutions that are patent protected or protectable. Second, within universities, huge overhead costs often make the exploration of commercialization expensive and time-consuming. Third, the size of the market might be too small to get sufficient return on investment to keep the business viable. Fourth, there is the risk that potential patent infringement is targeted aggressively by established companies as soon as patented technology leaves the realm of pure academic use. Fifth, there is often a lack of knowledge of academics in the areas of economics and experience of how a project is turned from a prototype into a product. Sixth, the reluctance of some academics to devote part of their time to setting up a business. Low-cost microscopes like the Foldscope and the associated worldwide dissemination of small microscopes in places where they would otherwise not be found can lead to discoveries that can be shared and discussed both in class and with the wider community, e.g., on social networks. The associated ease of access to these tools, which are available to virtually everyone, makes education more inclusive. These novel ways of education also support the growing interest in STEM subjects. Training young interdisciplinary professionals with the help of open-source tools promote and create international cooperation. An important element for the future is making the resources comprehensive to reduce the burden on the educators and provide the easiest possible access for direct use in the classroom. In the past, advanced optical light microscopy was seen as an expensive specialist endeavor. Open microscopy, similar to the open-source movement as a whole, breaks down barriers in microscopy at all levels. We are convinced that this applies just as much to cuttingedge microscopy-driven research as it does to applications of simpler microscopy methods in areas such as healthcare and education. For the large pool of microscopists for whom biological discovery is the key driver, the goal is not to apply the method with the best resolution on paper. Rather, the aim is to find or develop the most suitable technique, or combination of techniques, that work within the constraints of a specific biological question. These researchers benefit from the modular nature of open microscopy when they rapidly test, prototype, and tailor different microscopy approaches for their specific system, and often combine multiple advanced microscopy techniques in a way that simply would not be feasible in either commercial or traditional homebuilt systems. This allows researchers to use the best microscopy tool for their project, At its best open microscopy empowers scientific curiosity, creativity, and collaboration. For this reason alone, it is worth investing time and money into its bright future. A list of hardware and software projects, repositories and additional resources can be found on https://github.com/HohlbeinLab/OpenMicroscopy. The authors welcome contributions to make the list comprehensive and keep the list up to date. J.H. and K.P. initiated the manuscript. All authors provided sections and contributed to the final version of the text. Foldscope: Origami-Based Paper Microscope Evaluating the performance of a low-cost mobile phone attachable microscope in cervical cytology Robotic microscopy for everyone: the OpenFlexure microscope A versatile and customizable low-cost 3D-printed open standard for microscopic imaging ImJoy: an open-source computational platform for the deep learning era NIH Image to ImageJ: 25 years of image analysis Fiji: an open-source platform for biological-image analysis Democratising deep learning for microscopy with ZeroCostDL4Mic Characterization of an industry-grade CMOS camera well suited for single molecule localization microscopy -high performance super-resolution at low cost The FAIR Guiding Principles for scientific data management and stewardship Open Hardware: A Key for Accelerating Science and Technology Towards the U.n Sustainable Development Goals (sdgs) Haves and have nots must find a better way: The case for open scientific hardware The €100 lab: A 3D-printable open-source platform for fluorescence microscopy, optogenetics, and accurate temperature control during behaviour of zebrafish, Drosophila, and Caenorhabditis elegans A Framework for 3D Printable Optomechanics Open configurable high-throughput imaging platform for infectious disease diagnosis in the field The Field Guide to 3D Printing in Microscopy OpenSPIM: an open-access light-sheet microscopy platform The mesoSPIM initiative: open-source light-sheet microscopes for imaging cleared tissue Tutorial: practical considerations for tissue clearing and imaging Crossbill: an open access single objective light-sheet microscopy platform Nanometer-scale Multiplexed Super-Resolution Imaging with an Economic 3D-DNA-PAINT Microscope Visualisation of dCas9 target search in vivo using an openmicroscopy framework Laser-free super-resolution microscopy Squid: Simplifying Quantitative Imaging Platform Development and Deployment An Open-Source Framework for Automated High The smfBox is an open-source platform for single-molecule FRET Single-molecule detection on a portable 3D-printed microscope Cost-Effective, Open-Source Microscope for Scanning Two-Photon Microscopy that Is Modular and Readily Adaptable Video-rate multi-color structured illumination microscopy with simultaneous real-time reconstruction Cost-Effective Live Cell Structured Illumination Microscopy with Video-Rate Imaging Microscopes Using µManager Pycro-Manager: open-source software for customized and reproducible microscope control 3D Multicolor Nanoscopy at 10,000 Cells a Day Python-Microscope: High performance control of arbitrarily complex and scalable bespoke microscopes ImSwitch: Generalizing microscope control in Python EMU: reconfigurable graphical user interfaces for Micro-Manager Micro-Magellan: opensource, sample-adaptive, acquisition software for optical microscopy The Open Microscopy Environment: A Collaborative Data Modeling and Software Development Project for Biological Image Informatics Super-resolution fight club: assessment of 2D and 3D single-molecule localization microscopy software Standardisation of Practices in Open Source Hardware A perspective on Microscopy Metadata: data provenance and quality control QUAREP-LiMi: A community-driven initiative to establish guidelines for quality assessment and reproducibility for instruments and images in light microscopy PYMEVisualize: an open-source tool for exploring 3D super-resolution data Content-aware image restoration: pushing the limits of fluorescence microscopy Star-convex Polyhedra for 3D Object Detection and Segmentation in Microscopy Cellpose: a generalist algorithm for cellular segmentation QuPath: Open source software for digital pathology image analysis Real-time multi-view deconvolution Rapid image deconvolution and multiview fusion for optical microscopy CLIJ: GPU-accelerated image processing for everyone In Toto Imaging and Reconstruction of Post-Implantation Mouse Development at the Single-Cell Level An Open-Source Modular Framework for Automated Pipetting and Imaging Applications Circuit Investigations With Open-Source Miniaturized Microscopes: Past, Present and Future. Front Emerging Business Models for Open Source Hardware The authors thank their colleagues and SciTwitter for inspiration and discussions. We thank We thank Dimitrios Tsikritsis for providing valuable feedback to the manuscript.