July 30, 2018 – If you are a pharma or biotech company, a tissue bank, or a reference or specialty lab, how do you choose your tissue image analysis technology or service provider? Even if you were the world’s top technology expert, how could you evaluate a vendor who claims to do Deep Learning, a really cool technology everybody is now talking about? The dilemma is that it is not so much about the technology, but the data: DATA!= DATA. Just getting data or even a lot of data and having a cool technology doesn’t necessarily give you the right data. You cannot just compare technologies. Now, with immuno-oncology (IO) in need of tissue data, the stakes are high and getting the wrong tissue data can cost you a lot of money.
We have created a framework to identify the relevant core competencies of the different types of companies in this space, in the form of the spider plot shown above. Conceptually, we can identify 3 different types of companies:
- Histology labs using 3rd party image analysis tools for very simple needs (such as Her2, ER and PR),
- Image analysis software companies, who provide off-the-shelf image analysis and machine learning tools with some technical services to tune the tools to specific applications, and
- Full-service providers that can take care of everything from start-to-finish, all the way from staining the tissue to delivery of the data including its interpretation.
Here are a few things to consider when choosing who you trust to get you the right data:
- THE RIGHT TEAM. The key to getting the right data is to bring together the expertise of pathologists, biologists, image analysis & machine learning experts and a histology lab to develop a holistic solution, whether you have them in-house or as part of your service provider. Cross-discipline communication is the challenge, as different disciplines use different languages. Make sure you have a multi-disciplinary cross-trained team that can work together effectively.
- PROVEN TECHNOLOGY FOR THE REAL WORLD. Has the technology been developed by a pure technology company (engineers only) or academia, just because it is a cool technology? It takes real-world problems, and a lot of them, as well as that multi-disciplinary team who understands all facets of the problem to drive the right technology development. Has the technology been developed and demonstrated on just a few demo slides? The ability to use a technology in research does not mean it performs in a clinical setting. Medicine in the real-world is very complex! Has your technology solution already proven itself on thousands of slides and in a clinical setting?
- TECHNOLOGY THAT ADAPTS QUICKLY TO YOUR NEEDS. Your needs will evolve quickly! 5 years ago, nobody saw a need for tissue image analysis in Immuno-Oncology, it is amazing how quickly that has changed. New technologies or capabilities may be needed even during an ongoing project, as the understanding of the biology unfolds. How quickly can your technology solution keep up with your changing needs? Or if you are working on a rare disease, who is providing your technology? Off-the-self products provide mainstream solutions, have long development cycles, and may not ever consider your needs on their roadmap.
- WHO OWNS THE TECHNOLOGY? Competitors can be beneficial for your developments, or they can be a thread to your business. Working with a technology or service provider who works with companies across the entire industry allows you to benefit from the latest advances in technology that reflect the latest industry trends. Working with a technology or service provider that is owned by a competitor, puts you at risk of exposing your development strategy and might get you into an unwanted dependency.
- TECHNOLOGY IN THE RIGHT HANDS. Research and new applications require a lot of flexibility with the tools, and this means complexity. While pathologists and biologists have a good understanding of the final data they need from a science perspective, it takes a technology expert, who has a thorough understanding of the underlying technology, to handle the tools properly to get the right data. Do you, or your service provider, have science and technology cross-trained experts to operate the tools?
- THE RIGHT PRICE. As applications mature, many tissue image analysis applications become standard product offerings, with a clear interpretation and at a low cost, as they can be provided as a streamlined service. But caution: DATA != DATA. Low cost may also mean: limited technology, little effort/work or low quality.
- A SCALABLE TECHNOLOGY. You need high-throughput? Off-the-self tools intended for the research market are not necessarily designed to work in a high-throughput operation. Robustness, scalability (HPC backend) and support (engineers on staff) become important factors. A lot of burden will be placed on IT to make it work and keep it maintained. Has your tissue image analysis solution already proven itself in a high-throughput operation?
- A SCALABLE OPERATION. How scalable is your operation when it comes to tissue image analysis? Do you need overflow capabilities? When supporting clinical trials, resource management can be a challenge (uncertainty and lumpiness). Full-service providers could be the solution for you.
- COMMERCIAL PATH. Once you are getting the right tissue data in an exploratory setting, is there a continuous path to your clinical trials and then the clinical market? Do you have a lab, or does your service provider have a lab, that can run the technology (digital pathology + image analysis) in a regulated environment (CLIA, CAP, “FDA regulated” LDT, …)? Does your technology/service provider own the technology to be able to manufacture it as a medical device? Does your technology/service provider have a Quality Management System (QMS) in place to manufacturer medical devices and the appropriate regulatory expertise (IVD) to get regulatory approvals (e.g. FDA)?
Flagship learned the importance of those concerns from our customers and they defined the company we have become: the leading full-service provider for tissue image analysis with our own in-house technology development. With our core competencies in operations, science, technology, and regulatory we have created a blue print for a tissue image analysis operation that also scales nicely to multiple sites and internationally.
If you want to learn more about the technology behind tissue image analysis, see our LinkedIn article series , which reviews the different aspects of Pathology AI, intended for a broad non-technical audience.
Holger Lange, PhD
Chief Technology Officer
Flagship Biosciences