Vetology has rebuilt all 94 of its artificial intelligence radiology classifiers on a new model architecture. The updated models were rolled out between June and August. The new architecture allows the AI to interpret global information across single and linked images within a broader study, mirroring how a human radiologist evaluates cases.
The company noted that this development addresses a recent joint position statement from the American College of Veterinary Radiology and the European College of Veterinary Diagnostic Imaging. The position called for improved transparency, error reporting, and third-party evaluation of commercial veterinary AI services.
Anthony Lutz, Vetology’s manager of data science, stated in the release that the new models represent a “radical shift.” He noted that the AI’s ability to interpret linked images “follows how a human radiologist reads cases, and the success of these models flows directly from that design.”
To further build clinical trust and accountability, the company maintains an open invitation for independent researchers to test its published condition-level performance data.