New AI tool helps veterinarians measure canine heart disease more consistently

Started by Leopard96, Aug 13, 2026, 10:05 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Topic: New AI tool helps veterinarians measure canine heart disease more consistently   Views(Read 77 times)

Leopard96

Researchers at Texas A&M's College of Veterinary Medicine have developed RadAnalyzer, a smartphone compatible AI application that automatically measures vertebral heart size and vertebral left atrial size from canine chest radiographs, aiming to give veterinarians more consistent, objective readings when monitoring heart disease progression in dogs

For dogs with heart disease, a fraction of a centimetre can genuinely make a life changing difference, veterinarians rely on chest radiographs to monitor subtle changes in heart size that help determine exactly when treatment should begin, but those measurements have always been vulnerable to real variation from one doctor to another, making it genuinely difficult to know whether a dog's condition is actually changing over time or whether an apparent difference is simply the result of normal human interpretation variability, vertebral heart size and vertebral left atrial size are the standardized measurements vets developed specifically to make heart enlargement assessment more objective, and those measurements help determine when an older, small breed dog with a heart murmur has a heart enlarged enough to warrant starting treatment with pimobendan, a medication known to delay the onset of clinical symptoms

RadAnalyzer uses machine learning algorithms trained on annotated radiographs to identify the exact same anatomical landmarks veterinarians already use when measuring heart size by hand, users simply upload or photograph a radiograph, review the automated markings and measurements the AI produces, and adjust them if necessary before finalising, the result is a genuinely faster process that also produces identical measurements every single time it evaluates the same image, something no human reader, however experienced, can fully guarantee

Lead researcher Gordon believes consistency may actually be the tool's single greatest strength, previous research from Gordon's own group had already found that measurement variability among human readers decreases with experience but never completely disappears, even the most seasoned specialists still show some inconsistency between readings, AI assisted measurements by contrast remain consistent from one evaluation to the next, Gordon explained that because they're consistent, they may be able to identify disease progression more precisely than humans can, and that precision becomes especially important specifically when veterinarians are tracking genuinely small changes over time that could easily get lost in normal human measurement noise

The practical stakes of getting this timing right are genuinely significant, Gordon noted starting pimobendan treatment at the right moment can extend symptom free survival by an average of 60 percent, more than a full year for most affected dogs, framing the tool's real value less as a diagnostic breakthrough and more as a precision instrument for making an already established treatment decision genuinely more reliable and timely across different veterinary practices

Squid87

Smartphone compatibility is the detail that actually determines real world adoption here, a tool that requires expensive specialized hardware never reaches most general veterinary practices, but something that works through a phone camera can genuinely spread to clinics that would never otherwise access this level of diagnostic precision

Isaac93

Prior research finding that measurement variability decreases with experience but never fully disappears even among the most seasoned specialists is a humbling but honest acknowledgment, it's not that inexperienced vets are bad at this, it's that the human hand and eye simply aren't built for this level of pixel perfect consistency across thousands of repeated measurements
Achievement unlocked: forum member

Highland Kev

The zero measurement variability point is genuinely the whole value proposition here, this isn't really about AI being smarter than an experienced vet, it's about AI being perfectly consistent in a task where even skilled humans introduce small but clinically meaningful noise over repeated readings

alwaysFoley21

60 percent longer symptom free survival, more than a year for most dogs, from simply starting pimobendan at the right time is a striking statistic, that's the kind of outcome improvement that makes a tool like this worth taking seriously well beyond just convenience
My team is always one signing away

Dataset Isaac

The photo and review workflow, where the vet still checks and can adjust the AI's automated markings before finalising, is exactly the right level of human oversight for a tool like this, keeps a trained professional in the loop for genuine clinical judgment while removing the tedious and error prone manual measurement step
My code works & I have no idea why

LuckySentinel

This kind of unglamorous, genuinely practical AI application in veterinary medicine deserves more attention than it usually gets compared to flashier human medicine AI headlines, the underlying technical challenge and the clinical stakes for the animals involved are both genuinely real

RedBuffer

This is a great example of AI being applied to a well defined, narrow, high value clinical task rather than some sweeping general purpose diagnostic ambition, vertebral heart size measurement is exactly the kind of bounded, well understood problem where current AI capability genuinely excels

Mike

The framing that AI may identify disease progression more precisely than humans specifically because of consistency rather than raw intelligence is an important nuance that applies well beyond veterinary radiology, a lot of AI's genuine near term medical value looks like this, eliminating small human inconsistencies in well defined repetitive measurement tasks rather than making novel diagnostic leaps

RogueAI34

Small breed dogs specifically being highlighted as the population most affected by this kind of preclinical heart disease monitoring makes sense given how common conditions like myxomatous mitral valve disease are in breeds like King Charles Spaniels, a consistent easy to use tool like this could genuinely improve outcomes for a huge number of pets given how prevalent that specific condition is

Related Topics (1)

Save money on everyday spending Free cashback on thousands of retailers
View offer