Founded By A Veterinarian, The Start-Up Is Developing An Artificial Intelligence-Based System For Radiological Diagnosis Of Horses

“Innovatio veterinariae,” a start-up founded by veterinarian Rugilė Dauliūtė last year, develops software for automating the evaluation of equine radiological images based on artificial intelligence (AI).

“To create innovations, Lithuanian start-ups actively use artificial intelligence that is rapidly gaining ground in the world. In radiology, artificial intelligence is used to detect and diagnose pathologies, because, thanks to deep learning, algorithms are already able to process large amounts of data, such as images. The growing number of innovators is a great proof that we are strong in the Life sciences and that we are not inferior to advanced countries in methods we develop and research we carry out. This undoubtedly affects not only the field of medicine but also the economy,” points out Gintas Kimtys, the acting director of MITA.

innovatioveterinariae.jpg (original, 852x536)

Indeed, orthopaedic disorders in horses are of great economic importance. When they get sick, not only does the value of the horse drop drastically, but the sporting career is also put on hold or even ends. One of the main diagnostic procedures for detection of orthopaedic equine diseases is radiological examination. It is used to determine causes of limping, loss of the athletic form, or in the case of purchase and sale of animals.

Rugilė Dauliūtė, who is currently doing her postgraduate internship at a racehorse clinic in the Pays de la Loire region of France, has been accompanied by horses since early childhood. From the age of ten, she was involved in show jumping, endurance riding, equestrian competitions, and later, trained young riders. While studying at the Veterinary Academy of Lithuanian University of Heath Sciences in Kaunas, for five years, Rugilė worked at the Large Animal Clinic.

Veterinarian Dauliūtė got the idea of a radiological diagnostic system in 2018. While still in practical training at horse clinics in France, Denmark, Germany, Hungary and Portugal, she found out that taking X-ray images well was only half of the job. Due to the size of the animal, 8 to 42 X-ray images are taken in one study; the interpretation and analysis of X-ray images is a complex part of the work that requires a lot of specific knowledge about the veterinarian’s job. For these reasons, pathologies are sometimes unnoticed and underestimated, although further operation of the horse depends on it – whether surgery is required and so on. By taking interest in the development of innovative products in her areas of interest, Dauliūtė learned about the artificial intelligence used in human medicine to solve other similar visual problems and decided to offer diagnostic assistance to veterinarians.

applyingaitoimageanalysis.jpg (376x240, 376x209)“Applying AI to image analysis will give veterinarians a “second eye” for determining diagnosis faster and more accurately and help to provide timely treatment. The system will be able to identify the metacarpophalangeal/metatarsophalangeal joint from new images and determine in which projection is the image taken. The area of this joint is the most vulnerable spot among sport horses. Also, the models will be able to identify two critical radiographic findings from new X-ray images – fractures and osteochondral fragmentation. It will take another 2-3 years for the product to be developed,” estimates Dauliūtė.

x-rayimage2.jpg (376x240, 376x190)A team assembled by the veterinarian is currently developing a prototype of software for automating the evaluation of equine radiological images: three IT professionals have already developed a platform and full internal system architecture for image processing, an artificial intelligence expert is developing model algorithms, 14 international students are involved in image categorization, and 10 veterinarians from the UK, Lithuania, Italy, Denmark and France mark X-ray photographs.

When asked about the common problems for those who work with AI systems, Dauliūtė pointed out that it is not easy to get “patient” data for system development. This is facilitated by the trust earned in internships and cooperation with foreign equine clinics and private veterinarians, who share their patient database. It is hoped that as the project gains momentum, it will be easier to obtain data.

The target market for the system under development is equine veterinary clinics and private veterinarians, and the end-user is a veterinarian. The future plan is to integrate the system with software from other companies for archiving and sharing images of diagnostic tools such as radiography, ultrasonography, magnetic resonance imaging, scintigraphy and computed tomography.

rugildaulit.jpg (original, 505x724)According to the veterinarian, only one company in California, USA is developing a similar solution in the horse market: “They have a system that automatically makes various measurements of horse hooves used by blacksmiths and veterinarians. As the horse market is relatively small, we keep in touch with this company, exchange news and hopefully will combine products in the future. The other four companies developing algorithms with computer-aided diagnosis operate only in the small animal market.”

The team is developing the idea of an equine radiological diagnostic system based on artificial intelligence in various programs. They are currently not only developing a prototype but also trying to attract funding from various programs or startup accelerators. “We actively participate in pre-accelerators and exhibitions. However, it was the “Inostartas” project that prompted us to take the project to the next level – to establish a company, to employ people. We are currently participating in the mentoring program “Life Sciences Start-up MasterClass” of the project “Promotion of Life Sciences Industry Development” implemented by the agency, and we consult with experts of the project “AI Boost” on artificial intelligence issues,” concludes Dauliūtė.

Photos: Rugilė Dauliūtė
Prepared based on the data taken from the Agency for Science, Innovation and Technology