The Growing Role of Artificial Intelligence in Orthopedics

VinLab
4 min readMar 26, 2023

The World Health Organization estimates that about 1.7 billion people worldwide suffer from diseases and injuries related to the musculoskeletal system, which comprises locomotor organs like bones, joints, muscles, ligaments, and tendons. Orthopedics, the branch of surgery concerned with the musculoskeletal system, is therefore among the most discussed topics in healthcare. With the rapid development of information and data technology, clinical applications of artificial intelligence (AI) in orthopedics have been gaining lots of attention for relieving surgeons’ workload in numerous aspects.

Let’s look at how AI might help orthopedic doctors make better diagnoses, enhance treatment, and improve patient outcomes.

Diagnoses with AI

AI applications in orthopedics have been focused primarily on image interpretation for diagnostic purposes. Computer vision is a critical function application for such a task, and for this reason, convolutional neural networks are typically employed.

A study in 2021 by researchers at the VA Boston Healthcare System has developed an artificial intelligence system that can help radiologists auto-detect fractures in X-rays and highlight suspected regions. The AI system, called BoneView, was trained on 60,170 radiographs to recognize fractures of the limbs, pelvis, torso, lumbar spine, and rib cage. To evaluate the performance of AI-assisted medical image reading, 12 radiologists and emergency physicians were asked to identify and localize fractures on radiographs with and without BoneView’s assistance. The result is that with the help of AI, missed fractures were reduced by 29%, reader’s sensitivity and specificity improved by 16% and 5% respectively, and reading time per patient went down by 15%.

AI applications like BoneView are powerful tools to help radiologists and doctors to improve diagnostic accuracy and efficiency, while potentially enhancing patient experience during their therapeutic pathway. AI-powered assistance could reduce the chance of undetected damages in emergency settings, including cases where patients may sustain multiple fractures. It also reduces workload and speeds up the workflow for medical professionals in emergency departments, allowing them to make timely assessments of patient conditions. Moreover, AI can serve as an educational tool for physicians, reassuring them when their diagnosis is correct or prompting them to reassess before treating patients.

Surgery with AI

AI is considered to have game-changing impacts on all stages of the operation process. One of the recent applications of AI in this field is surgery simulation, which is an increasingly popular tool for enriching the hands-on experience of surgeons.

Manufacturers of these AI systems are trying their best to incorporate all elements of the operation room into their virtual reality simulators. Swiss company Virtamed provides medical training simulators that include custom-designed surgical tools, scalpels, drills, and saws. Supporting orthopedic, ob/gyn, laparoscopic, and urological operations, this tool helps enrich surgeons’ training experience and build their muscle memory with tools, all in a highly realistic anatomical environment. Training on these simulators can promote surgeon competence and confidence without having to practice on real patients.

Canadian simulator manufacturer Symgery takes training to a new level by providing realistic 3D surgical environments. Surgeons can refine a wide range of skills as they replicate surgery techniques and procedures on simulators that mimic real-life clinical scenarios. To enhance the authenticity of the experience, the company even supplies AI-generated haptic feedback following users’ actions on the model. The training experience is complemented by real-time personal feedback and the option to stop halfway and restart. Thus, simulation provides wonderful educational opportunities for surgeons to observe scenarios that could play out in a real surgery and learn how to navigate them without having to compromise the health of real patients.

Limitations of AI

Although AI has empowered many accomplishments in the orthopedic field, it is still a new technology that has not been thoroughly explored. Therefore, it is important to keep the limitations of AI applications in consideration in order to accurately evaluate the possibilities and properly integrate them into the future healthcare system.

  • Responsibility: There is still much dispute over who should be given responsibility in the case that AI produces miscalculations. Therefore, most AI-powered medical assistants are still operating under the supervision of a medical professional who verifies and approves AI’s predictions before disclosing them to patients.
  • Authenticity of experience: Many are worried that simulation can never replicate the atmosphere, emotions, and pressure present in a real operation. However, it should be noted that simulations are for educational purposes only, allowing junior surgeons to observe and make mistakes that they are otherwise not permitted to make in real-world situations. Besides simulators, textbook knowledge and guidance from senior surgeons are always important components throughout their learning journey.
  • Expensive adoption: Currently, AI medical assistants in orthopedics (and other medical branches) are still considered expensive for mass adoption. It is estimated that the cost of AI solutions in medicine ranges widely from $20,000 to $2 million, so currently, private clinics and remote medical centers are mostly missing out on the benefits of AI.

Thanks for reading!

If you are looking for information about artificial intelligence, machine learning, general data concepts, or medical data science applications, follow us to acquire more useful knowledge about these topics.

Contact

Email: info@vinlab.io

Twitter: https://twitter.com/VinLab_io

YouTube: https://www.youtube.com/@Vinlab-MedicalImageAnnotation

Open source project: https://github.com/vinbigdata-medical/vindr-lab

--

--

VinLab

A Data Platform for Medical AI that enables building high-quality datasets and algorithms with lean process and advanced annotation features.