In the summer of 2025, I had the privilege of working in my school's research lab as a fellow. One of the professors there was leading a project that relied heavily on microscopy, and her research fellows produced textbook-beautiful tissue slides. I'd often peek over at their displays just to see what they were examining.
While collecting stock data for Eosin, I casually asked whether we could prepare a few slides from my own buccal swab. To my surprise, her microscope—the same instrument responsible for some of the most beautiful slides I'd ever seen—couldn't perform whole slide imaging at all. Looking into it further, I realized that whole slide imaging is surprisingly uncommon, even among well-funded research labs doing serious, high-impact work. The best we can do is stitching individual tiles together—a tedious process that inevitably confers artifacts without a CNC machine and automation. Regrettably, we don't have a CNC machine in the lab.
The disconnect became even more obvious after digging into the digital pathology landscape. While researching similar projects, I looked into every startup, academic tool, and "next-generation" platform I could find. Almost all of them were closed-source, opaque, and built around glossy marketing pages. Open source options are abandoned, don't scale horizontally, or boast problematic licenses.
This is why I'm open-sourcing Eosin. If computational pathology is going to move forward, the basic infrastructure needs to be transparent, inspectable, and available to anyone who wants to build on top of it. Eosin is released under a permissive MIT/Apache 2.0 dual license so developers, researchers, institutions, and even businesses can use it freely, modify it, and integrate it into their own pipelines without legal friction. This dual-license approach is widely considered one of the most permissive and enterprise-friendly.
Eosin isn't meant to be a competitor. It's meant to be a foundation—an accessible, high-performance starting point for anyone working with whole slide imaging. By making the stack publicly available, I'm hoping to lower the barrier for students, researchers, and engineers who want to experiment with computational pathology without needing a grant, a sales contract, or a proprietary pipeline.
Open sourcing a viewer is just a starting point. The deeper issue is that pathology lacks a modern computational foundation altogether. Most labs are still structured around manual workflows, proprietary hardware, and file formats that don't lend themselves to automation or analysis. Even when slides are digitized, the surrounding infrastructure—storage, retrieval, annotation, search, quality control—remains fragmented or entirely missing.
I believe pathology needs a unified substrate: an architecture that treats whole slide imaging, automation, and AI as first-class citizens rather than afterthoughts. Eosin is my attempt to take a small step in that direction by providing an open, modern, and extensible base layer.
If you're a developer or engineer who finds this space interesting, I'd be glad to have you involved. Eosin is open source partly because I want it to be useful, and partly because I think pathology should have the same kind of shared, well-engineered tooling that other fields take for granted.
You don't have to be in medicine to make a meaningful contribution. Even small improvements can help clinicians, students, and researchers who come to depend on these tools. If nothing else, it's a chance to explore a part of software engineering that most people never see, and one that still has a lot of room for innovation.
Here's the link to Eosin's source code: github.com/eosin-platform/eosin
Whether you contribute code, ideas, or feedback, you'd be welcome.
-Tom