In recent years, the veterinary world has witnessed a technological revolution propelled by artificial intelligence (AI). As of the end of 2023, AI has permeated various professions, raising concerns among traditionalists about its potential takeover of routine and essential tasks. However, the question arises: Is AI in veterinary radiology a force to be feared and shunned, or should it be embraced and fully utilized?
Research from Tobias Schwarz, MA Dr. Med Vet, FRSB, PgCAP, FHEA, DipECVDI, DACVR, DVR, FRCVS, of the University of Edinburgh’s Royal (Dick) School of Veterinary Studies in Scotland sheds light on this debate, suggesting veterinary radiology AI is not intended to replace radiologists, but rather to aid and complement their work.1,2 This perspective aligns with my firsthand experience, witnessing how AI technology is designed to support veterinary teams, reduce their workload, offer peace of mind, lower costs, and, most crucially, enhance the speed and accuracy of care—a particularly valuable contribution given the high demand for veterinary radiologists.
According to Daniel Levenson, DVM, owner of two small animal hospitals in Albuquerque, N. Mex., “This study shows promise in veterinary AI technology’s ability to improve the efficiency and quality of care.”1
Between packed schedules and comprehensive reports, the essential work of veterinary radiologists is often urgent and can, in many cases, make a life-saving difference to pets in critical condition. Still, as many radiologists acknowledge, the system is far from perfect. With a naturally demanding profession, these experts are all too often overworked and could greatly benefit from additional resources that help make their jobs more efficient.
“This technology is a real breakthrough for clinics that are understaffed or lack a dedicated expert in veterinary radiology. In these cases, a patient’s outcome could depend on a rapid report or even a second opinion,” Dr. Levenson says.
Pets’ lives often depend on having the highest-quality and most advanced technology at their veterinarian’s disposal to enable accurate results, delivered as fast as possible. Leveraging advanced AI tools is an excellent opportunity to modernize veterinary technology, providing an objective and constantly learning resource for veterinary teams when they need it the most.
As the first generations of veterinary radiology AI software become commercially available, it is essential to evaluate their accuracy compared to seasoned veterinary radiologists. Dr. Schwarz and his team have conducted a study comparing radiological interpretations by veterinary radiologists with those generated by a veterinary radiology AI software. The study, involving 50 canine and feline cases, aimed to assess the software’s performance in X-ray interpretation reporting.1-3
Utilizing technology that employs machine learning to assess radiographs in real-time, the researchers evaluated the AI’s accuracy based on sensitivity, specificity, and overall reading accuracy. The findings were compared to the interpretations of human radiologists, with the correct reading determined by the consensus of the majority. The study yielded results with AI outperforming a number of radiologists and showing no significant difference in accuracy compared to the highest-performing radiologist.1
Additionally, the software used in the study has a bias intended to support the most careful diagnosis of the pet behind the images. Although the AI was biased toward claiming abnormal results, especially with unclear findings, the AI still performed almost as well as the highest-performing radiologist in all settings for radiographic findings.1- 3
The research suggests the current technology excels in identifying clear cases of abnormality and providing a second opinion on ambiguous or potentially abnormal results.1 Rather than posing a threat to veterinary radiologists, the AI acts as an ally, guiding professionals to scrutinize complex cases more thoroughly. This collaborative relationship between AI and radiologists positions the technology as an asset in enhancing the diagnostic process.
However, it is essential to acknowledge the study’s limitations. Schwarz’s research primarily assessed single cases, lacking a comprehensive analysis of the animals’ medical history—a crucial aspect of radiological interpretation. Real-world scenarios involve considering an animal’s health holistically, and future studies should explore AI’s ability to analyze medical histories and compare radiographic results.1
The outcome of this particular study’s findings suggests the current state of veterinary radiology AI is not a flawless replacement but a complement to the work of skilled imaging specialists. As AI technology evolves, it holds the promise of revolutionizing veterinary care, offering enhanced diagnostic capabilities, and ultimately improving health outcomes for pets. Embracing AI as a valuable tool in its early stages may redefine the landscape of veterinary radiology and set a precedent for future advancements in the field.
As AI continues to advance, learning from its users and refining its precision, the potential for significant improvements in veterinary care becomes evident. This technology contributes to its ongoing enhancement, setting a new standard for veterinary care. One valuable consideration is the AI’s ability to rapidly improve based on the results of veterinary teams globally. So, without necessarily intending to, vets using this technology are helping provide their fellow colleagues using this same technology all around the world with the most accurate results possible.
The usefulness of the current available technology is clear. It is a valuable resource to enhance the work of the professionals bold enough to accept advanced technology, become early adopters, and start setting a new standard for what cutting-edge veterinary care really looks like.
Avi Avner BVSC, CVR, DVDI, MRCVS, is a radiologist specializing in veterinary diagnostic imaging. Avner is also an imaging consultant for pre-clinical and clinical studies, with more than 15 years of experience. In addition, he is a consultant for multiple medical innovation projects, such as trans-catheter cardiac annuloplasty and repair, valve replacement, imaging-guided medical augmentation, and ultrasound-guided intravascular device implantation. Avi currently works as a freelance medical imaging specialist at Knowledge Farm Imaging Specialist, a company in Zichron Yaakov, Israel. He has a special interest in animal welfare and medical innovation. You can reach him at aviavner7@gmail.com.
References
- Schwarz T, Ndiaye YS, Chernev C, Ockenfels AS, Crampton P (2023). Comparison of radiological interpretation made by veterinary radiologists and AI software for canine and feline radiographic studies.
- ACVR Annual Scientific Meeting, New Orleans, LA, USA, October 25 – 28, 2023. (oral presentation) abstract in: Proceedings; Veterinary Radiology & Ultrasound in press.
- SignalPET, “Establishing the Gold Standard: Comparing Veterinary Radiologists to Pet Radiology AI,” web log, SignalPET (blog) (SignalPET, December 13, 2023), https://www.signalpet.com/science/veterinary-radiologist-and-veterinary-radiology-ai-comparison-research/.