Today’s post is guest written by Dr. Steven Potts, MBA, CEO of Flagship Biosciences. My many thanks to Steve for his thoughtful commentary and personal thoughts on the meeting. Comments on your experiences are welcome if you attended the meeting.
As someone who enjoys fishing high mountain streams and rivers, a great place to find fish is at the convergence of multiple streams, where the interacting currents churn new nutrients and food up to the surface. Pathology Informatics combined three streams — Advancing Information Management in the Clinical Laboratory (AIMCL) and Advancing Pathology Informatics, Imaging and the Internet (APIII), which merged upstream previously, and Lab InfoTech Summit, a LIS oriented conference. The convergence was brilliant, and brought new ideas into this event.
For North American adopters of digital pathology, there are now two leading conferences, Pathology Visions in San Diego and Pathology Informatics. Pathology Informatics was clearly a product of years of collaboration, dialog, and competition in the Academy, an academic-led conference. It is likely that Path Visions will be driven by industry and Path Informatics by the Academy. What a great time for both conferences, and it will be good for the pathology community to continue to have distinctions between them.
Digital pathology and whole slide imaging was a key part of the conference, but not the central focus. This is also healthy, the goal is a fully digital, pathology laboratory, not digital pathology by itself. Whole slide images are part of the lab, but must fit into other technologies covered at this conference, like LIS integration, integrated databases, workflow, sample tracking, and integration with radiology.
My conference highlight was the Histopathology Image Analysis (HIMA) session on Sunday afternoon. It was a packed eight hours of leading innovators in pathology image analysis. The churning of new ideas by the multiple university collaborations in computer science and pathology departments are clearly bearing fruit. The bottleneck is larger annotated sample sets, but impressive inroads are being made on new pattern recognition techniques on hard problems (automated Gleason scoring will cause any image analysis scientist to go running back to the IHC ER/PR/HER2 refuge for cover). The continuing need for these exciting techniques is for validation on larger datasets, and this will require vision and funding to produce these large databases (are you listening, granting agencies?).
It was also great to see more usage of the technologies in pharmaceutical clinical trials, as pharma/biotech will continue to be the playground where new companion diagnostics tests are developed. It was impressive to hear of multiple academic medical centers collaborating in pathology and computer science. The new tests will not likely be as “easy” as ER/PR/HER2, so it is great to see many innovative image analysis ideas. Unfortunately, despite the conference being very well attended, and located near Cambridge, the current biotech center of the world, there were few pharma image analysis attendees. But this is not the fault of the content nor program, people have to find out they missed the party, to motivate attendance the following year.
There were a number of exciting software innovations by university researchers, ranging from image analysis to QA algorithms for scanning quality. I would encourage the community to think about the distribution of their ideas, as the first thing we want to do when reading a image analysis publication is to try out the technique on our own slides. A number of vendors now have software development kits that allow extensions and other approaches for getting technology into the hands of the users. Whatever route is taken, allowing users to be able to access the software is as important as the publication itself. This is what made academic gene/protein sequence bioinformatics so powerful was the ability to get the technology quickly into the hands of the users.
The electronic poster presentation sessions were well-done, and covered a broad range of topics. Not having to print a poster at Kinko’s late at night is a definite personal benefit, although the interaction of a powerpoint presentation to a large audience is certainly much different from a one-on-one discussion in front of a large paper poster. I see a parallel to moving from glass to digital slides in pathology presentations, certainly makes life easier, but it does fundamentally change the communication dynamics.
This conference represented a major step forward in pathology convergence, particularly between pathology and computer science. The rivers from other modalities, including radiology and cytometry (!) can wait until later years downstream for integration. My personal opinion is that a fully digital, pathology lab will be a bigger, more immediate benefit to the patient than the pathology labs sucked (or suckered) into radiology PACS systems.
My compliments to all involved with Pathology Informatics. I look forward to next year!