05/11/2026
At Fur & Feather Veterinary Hospital, we use real board certified radiologists for confirmation of our radiograph studies. You not only get initial interpretation from our doctors, but also the specialty doctors as well for the most comprehensive analysis. AI may eventually catch up to human brains but for now, we will utilize more reliable humans.
In a recent pilot study published in JAVMA, researchers evaluated six commercial veterinary radiology AI platforms using canine abdominal radiographs obtained from general practice cases with confirmed diagnoses. Overall diagnostic performance was found to be “low to moderate,” with notable variability between platforms.
One particularly illustrative case involved a dog with a surgically confirmed small intestinal foreign body caused by a river stone. While one AI system classified the study as normal, another identified the case as abnormal but failed to detect the foreign body itself, instead incorrectly diagnosing splenomegaly, abdominal mass effect, gastric foreign body, gastric dilation, and colonic distension.
Importantly, the study also found that sensitivity for detecting small intestinal obstruction — a potentially life-threatening condition — ranged from only 23% to 69% across platforms.
The authors concluded that, in their current stage of development, these AI systems are not yet suitable for autonomous interpretation of abdominal radiographs in clinical practice.
In an interview discussing the study, veterinary radiologist Dr. Stephen Joslyn stated:
“It’s ticking a dopamine box for them. They don’t know how bad it is, and they feel like it’s helping. It’s the blind leading the blind, unfortunately, at this stage.”
AI may eventually become a valuable adjunct in veterinary medicine. However, studies like this reinforce an important principle: radiographic interpretation still requires clinical judgment, contextual assessment, and confirmatory diagnostics.
Study cited in the comments:
