Charles River speeds AI pathology workflow

Pathology is one of those hidden bottlenecks in drug development. Before a toxicology study can really close, somebody has to review tissue slides, check image quality, confirm the right structures are present, and decide whether a finding is real or noise. That work is expert-heavy and slow. Charles River says it has now pushed much more of that chain into one AI-enabled digital workflow — and the company is putting real numbers on the payoff. ### What actually changed? On May 11, Charles River announced what it describes as an end-to-end digital pathology workflow for nonclinical research. The point is not one standalone AI model. It is the full stack — histology tracking, slide scanning, automated quality control, digital primary review, and AI decision support for pathologists — tied together inside the same process. (biospace.com) ### Why is pathology such a slow step? Because a lot of the delay happens before a pathologist even starts reading. Technicians still have to catch blurry scans, striping, folds, bubbles, tears, chatter, and other artifacts that can make a slide unreliable. If that check is manual, every study piles up behind a very specialized queue. Charles River is aiming its automation at exactly that choke point. (biospace.com) ### What is the AI doing here? First, it is doing slide quality control. Charles River says scanning QC is already in production and can automatically flag out-of-focus images, striping, and similar scan artifacts. Next come histology artifact QC, targeted for Q3 2026, and organ sub-compartment completeness QC — meant to verify required organ substructures are actually present — targeted for Q1 2027. That last piece is the unusual one. It is less “is this image blurry?” and more “is the biology you need even on the slide?” (pharmiweb.com) ### What is already live besides QC? Charles River says GLP-validated digital primary pathology reviews are already implemented. That matters because digital pathology is only useful at scale if it can stand up in regulated preclinical work, not just internal experiments. The company says those validated digital reviews have already reduced pathology read times by an average of 20% by removing the need to physically ship glass slides. (biospace.com) ### Where does Deciphex fit in? Deciphex is the pathology AI partner under the hood for several pieces of this push. Charles River and Deciphex launched the Patholytix Foresight decision-support tool in April 2024, then expanded their partnership in February 2025 to build out integrated image management and additional AI tools for toxicologic pathology. So this week’s announcement is less a surprise launch than a broader packaging of capabilities the two companies have been assembling in stages. (biospace.com) ### How big are the promised gains? Charles River says the combined workflow can cut at least one week from standard pathology timelines. It also says AI-powered decision support and anomaly detection could improve pathologist efficiency by more than 30% once deployed more fully. The catch is that these are company claims from a product announcement, not a head-to-head published validation study. But even the narrower number — 20% faster digital reads already in place — is meaningful in a workflow where studies often wait on specialist review. (deciphex.com) ### Why does GLP language matter so much? Because in preclinical safety work, “faster” is not enough. The workflow has to be defensible, traceable, and acceptable in regulated environments. Charles River says it is developing GLP validation plans aligned with FDA guidance, and its digital pathology services page also points to 21 CFR Part 11-compliant image analysis capabilities. Basically, the company is trying to sell this as production infrastructure, not a flashy AI overlay. (biospace.com) ### Does this change anything beyond speed? Potentially, yes. Charles River is also tying the workflow to reduction of animal use where scientifically appropriate. The logic is straightforward — if pathology data moves faster and with fewer repeats caused by QC failures, studies can become more efficient upstream too. That is still an inference from the workflow design, but it fits the company’s stated positioning. (biospace.com) ### Bottom line This is really a workflow story, not just an AI story. Charles River is taking a part of drug development that still depends on manual checks and physical slide handling, then trying to turn it into a regulated digital pipeline. If the company’s timing and efficiency claims hold up in routine use, pathology stops being the quiet delay at the end of the study and starts looking more like software. (financialcontent.com)