Yes, another blog. After 10 years of building out Pathomation, a pioneering company in digital pathology, I’ve decided it’s time for something else.

The result? A new boutique consulting company called Pixel Path, which is tongue in cheek for “pixel pathway”: a phrase I heard once coined by a consultant during a customer meeting. The “path” suffix also refers to my original work in (digital) pathology, meaning that I’m available for consulting and assignments in that area, still, too. Last but not least: I did my dissertation work on the integration of “cellular pathways” in systems biology. It seems I’ve come full circle.

Even as a proverbial “one man show”, Pixel Path is a company in its own right. And every company needs a mission. Going by the state of current affairs in technology, I’ve decided Pixel Path should operate at the intersection of innovation, data science, and cloud computing.

Data science

This is an obvious one, or isn’t it? Why go for data science and not AI, the fashionable buzzword du jour?

I’ve been doing data science long before it was called data science. Back in 2001, I started working with Christi Magrath, a professor at Troy State University. She had downloaded a dataset from https://www.yeastgenome.org that she was trying to make sense of. It was a CSV file with about 1.5 million records. Not impressive in today’s context, but back then we’re talking about Microsoft Excel and a limit of 32,768 records.

So my first data science project was simple from a technical point of view: the data was convert to a Microsoft Access database, and SQL queries were run against it. I think a VB6 front-end application may have been involved as well to make it easier for her and her graduate students to interact with the data (molecular biologists typically don’t take up coursework in data modeling; pathologists don’t either for that matter).

I’ve been involved in a lot of data science over the years. My graduate work at Iowa State University involved organizing and integrating data at different resolutions from heterogenous system biology data sources. Hierarchical NetWork Integration (HINWI) was based on mathematical graph theory, with an intuitive front-end user interface again (MetNetDB). Oh, and there was a programmer-friendly back-end API, too.

MetNetDB was also my first foray into gigapixel images. Seamless zooming, HTML5… None of these things were commonplace yet in 2010: Think Flash, Java (applets), and Silverlight. My approach provided 3 pre-generated representations of each pathway, each at a different pixel-resolution. To some extend I was lucky: The images were still relatively small enough so they didn’t need something quite as complex as a tile server yet.

Forward to my recent adventure in (digital) pathology, and we’re talking about true gigapixel images, with many different use cases, any number of stakeholders, and complex content delivery requirements (hence the “pixel pathway” terminology). Specific use cases are available aplenty at my previous blog at https://realdata.pathomation.com.

Cloud computing

Ok, here’s one buzzword at least. But it’s an old and established one! I’m not bragging here about IoT, edge computing, DataBricks, or even Azure Redis.

I’ve been following cloud technology from the sidelines for the last ten years. When I first introduced the topic as an area to  keep an eye on, a discussion ensued where my partners were attempting to come up with their own definition of cloud computing. Lesson learned: cloud can mean many things for different people.

Why choose Pixel Path as a cloud-partner? Because apart from anecdotes and certifications, there are also facts and experience. At Pathomation, we got stuck somewhat halfway between “lift-and-shift” and true “cloud native”. From its early days, we allowed raw data to be stored in S3 containers. We appropriated EC2 VMs, and siphoned off relational data to RDS.

Then, we repeated the exercise in Microsoft Azure: S3 data was migrated to Gen2 data lakes, and RDS tables moved to Azure SQL resources. We started looking into Docker containerization as well.

Innovation

Even more than cloud computing, innovation means many things to many people. I’ve personally prefer the definition offered by Ideo, as it’s very concrete and actionable. Their framework is based on the premise that innovation takes place at the intersection of desirability, feasibility, and viability.

When you can identify something that people need, use existing technology to address the need, and then build a viable and sustainable revenue model on top of that: innovation accomplished.

Conclusion

I’ve created Pixel Path as a boutique consultancy shop for everything involving data science, cloud computing, and innovation. I have a successful and proven track record in life sciences.

For honest, experience-based assessments, cloud migration and PoC studies, Pixel Path is the place to go.

Convinced yet? Set up your first free consult with Pixel Path through Calendly.

Disclaimer: no part of this blog was generated by ChatGPT, but I make no promises towards to future (it’s just so easy to use when you just quickly want to have something explained or elaborated upon, but are at a loss for words yourself).