Helping biologists discover and distribute bioimage analysis workflows
Takeaways from user experience research on the new napari hub
We want to make it easier for biologists to analyze microscopy imaging data and to access emerging methods for bioimage analysis that leverage machine learning. To that end, for the past year, the CZI imaging tech team has partnered with the napari image viewer to enable high performance visualization and exploration of a broad range of imaging data. We’ve also conducted foundational user experience research studies to better understand the community we’re building for.
Creating a hub for plugins
For our latest project, we are building a site that biologists can rely on to find quality reproducible bioimage analysis workflows that are compatible with napari. It is designed to be easy to use and contribute to, allowing developers to efficiently share plugins with biologists, who then use plugins to analyze their imaging data. The imaging analysis ecosystem needs three core components to thrive: the plugins themselves, services to help build them, and a reliable mechanism to find and evaluate them. We’re working with a group of developers to support translating their segmentation plugins for napari, and our engineering team is hard at work on a suite of developer tools, but the challenge remains of a place to share these plugins with the broader community.
There are many ways to learn about a plugin today, from social channels to publication review, but a unified experience to support biologists in their quest for the right method, or ways to measure and improve the impact of a given plugin, still evades the community. napari plugins are currently distributed through PyPI; however, we felt there was an opportunity to offer a more dynamic user experience to cater to the full journey of plugin search, discovery, and installation.
To meet these varied needs, our team asked: how might we shift the plugin search experience from what our community described as “the wild west” into a self-reinforcing cycle of plugin distribution and discovery? To do so, we plan to create the napari hub: a center of activity that brings people together to exchange analysis methods, and a place that can grow with the needs of our imaging community.
We used a variety of user experience research methods to make the best informed choices about our first iteration of the napari hub, learning with and from the imaging community. Competitive analysis helped us understand the broader best practices of existing discovery and distribution sites; focus groups helped us understand how a successful service should empower users; and prioritization exercises created forcing functions to scope features most urgent and important to our first release.
Competitive analysis research is not a competition, but rather a way to learn from products that offer a similar service in order to build a mental model of the user journey and incorporate best practices into your own product. Our team looked at three types of distribution/discovery sites:
- Direct (other imaging plugin sites);
- Associated (sites that an imaging community member might use for another part of their work); and
- Indirect (sites that distribute/discover plugin-like entities that have nothing to do with imaging; for example, the iPhone app store).
While visual and feature choices vary between products, a core flow from the landing page to search results to the plugin page is vital. Successful sites welcome users with concise calls to action based on their needs, allow scoping via filter and sort, and display plugin details in a clean layout that allows them to decide whether they want to install it. Developers of plugins are offered tutorials and templates to ensure their plugin is successfully published. With these learnings, our design team was able to jump start on creating a user experience that relies on best practices while customizing the hub to the needs of the imaging community.
Focus Group Discussions
While the competitive analysis was helpful to understand site flow, focus groups played a vital role in including many voices to gain a better understanding of how the hub should support the community. We held five, 90-minute virtual sessions (thanks digital whiteboard tools!) with research biologists, imaging scientists, and current and future napari plugin developers and discussed what a perfect version of the hub would allow them to accomplish.
This conversation helped us define long-term value propositions to use as our north star. While open source software takes time and effort to build, these user stories will guide us through our first release and future iterations.
From our competitive analysis and focus group discussions, the team identified an extensive list of “must have” features for the napari hub. In order to prioritize the most urgent and vital features to offer in our first release, we ran an exercise called “buy a feature” with our focus groups: we assigned a “price” for each feature relative to its engineering lift, provided the group a “budget”, and asked them to negotiate between themselves to “purchase” a list of features to create the best first release of the napari hub. The exercise resulted in 22 distinct feature requests, though nearly all groups included the following:
- Filter and sort capabilities,
- Global search bar,
- Manual input of plugin description, including support links and citation guidance, and
- One-click installation.
The “shopping” experience helped not just in scoping features for our first release, but also fostered an invaluable conversation about future functionality. Some groups pointed out that helping biologists create “bounties” for new plugin requests will help connect them to developers to ensure they’re making plugins that are most urgent and vital to the community. Other groups said that helping developers manage and share machine learning models will bolster buy-in and help biologists understand the plugin’s intent. We are grateful to the community members who participated in these focus groups; our work would not be possible without their expert guidance. We look forward to investigating and building out a dynamic hub that scales and supports the community’s long-term needs.
From Research to Reality
With the support of community learnings, the imaging team is now hard at work building our first release of the napari hub. We are excited to debut this core part of our image analysis strategy and support a growing Python plugin ecosystem. As with all our work, we’re focused on staying close to the real problems, building for the long term, and using collaboration as a key tool to accomplish more together.
We hope the napari hub will offer immediate value to the imaging community by allowing users to see all available napari plugins in one site, search and scope plugin options across dimensions, and evaluate key metadata to make a more informed install decision. We’re continuing to work with the community to incorporate feedback and iterate often, adding important features like usage metrics and educational resources to help biologists more efficiently find the plugin that works for them, and help developers measure the impact of their plugin.
If you’d like to stay up to date on the napari hub, visit the napari website. Want to add your voice and participate in future user experience research studies? Sign up, and we’ll reach out when new opportunities arise.