HomeTechnologyArtificial intelligenceAI and hip preservation - Boston Children's Answers

AI and hip preservation – Boston Children’s Answers

VirtualHip uses artificial intelligence to convert data from clinical notes and radiological images into a comprehensive picture of a patient’s hip. (Illustration: David Chrisom, Boston Children’s)

Orthopedic surgeons and biomedical engineers are trained to approach adolescents and young adults hip pain from two different perspectives. Surgeons usually look for conditions such as femoroacetabular impingement (FAI) and hip dysplasia from a clinical point of view. Engineers are more likely to focus on the technological angle.

These two perspectives have come together at Boston Children’s Hospital, resulting in a tool that could improve diagnosis and clinical planning for hip patients around the world.

Two decades of hip data made accessible with AI

VirtualHip is a software platform that uses artificial intelligence (AI) and 3D imaging to support the diagnosis and treatment of hip abnormalities in children. The idea originated from conversations between Dr Ata Kiapourdirector of Boston Children’s Musculoskeletal Informatics Groupand Dr. Young Jo Kimdirector of the Hip preservation program in children and adultsthat started in 2017.

“I started following the surgeons on the hip conservation team to understand how they worked and how technology could help,” says Kiapour. In doing so, he discovered an unmet clinical need for personalized diagnosis and treatment planning. He also discovered a largely untapped source of data: images and clinical notes from more than 10,000 hip patients over the past 20 years.

Our goal is to provide detailed information so surgeons can plan the best treatment for each patient.”

The potential of such data is comparable to that of a doctor who has treated thousands of hip defects and remembers them all in great detail, says Kiapour. “That doctor can look at a patient’s clinical profile and tell you, ‘We’ve treated patients this way and found that this approach works best.'”

Must Read
Vision AI Startup Digs Into Digital Twins for Mining and Construction

But because the historical patient data was not searchable, it was largely inaccessible. To unlock its potential, VirtualHip uses AI to process historical clinical data and match current patients with past patients with similar clinical profiles. These historical, anonymized patient records will provide important insights as surgeons diagnose and plan treatments for their current patients. That is one aspect of VirtualHip. The other aspect involves extensive 3D image analysis.

The combined power of AI and 3D modelling

To diagnose hip abnormalities, orthopedic surgeons typically rely on a physical exam and diagnostic imaging. While some hip abnormalities are easily visible via MRI or CT scanning, many are not. Furthermore, the impact of some hip abnormalities often depends on the position of the hip. For example, in the case of FAI, a hip may only be painful when it is moving and the bones are rubbing together.

The limitations of traditional diagnostic imaging also complicate treatment decisions. Can physiotherapy solve the problem or does the patient need surgery? In the case of impingement, how much bone must be removed to allow the hip to move freely and without pain? In the case of dysplasia, at what angle should the surgeon rotate the acetabulum (hip socket) so that the femoral head fits tightly in the socket?

VirtualHip uses MRI or CT imaging to create a 3D model of the hip joint and then uses this model to automatically generate a comprehensive set of measurements of the hip structure. This data, combined with information from the clinical exam, provides a comprehensive view of the hip and areas of impingement or instability. Surgeons will be able to pinpoint the location and severity of the deformity with a level of accuracy not possible today.

A color-coded 3D image of a hip joint indicating the degree of hip dysplasia.
A 3D image of hip dysplasia in VirtualHip shows how much of the femoral head is covered (and not covered) by the acetabulum.
A simulated image of hip impingement.
A 3D image of hip impingement in VirtualHip shows where friction occurs when the hip is in motion.

“Once you have a 3D model, you can ask the software to simulate the full range of motion,” Kiapour explains. “These are the complex patterns that aren’t apparent on 2D images.”

Must Read
How to Succeed with AI Compliance in Manufacturing: Trustworthy AI for the Win?

With a detailed analysis of the hip’s structure and range of motion, VirtualHip will match the analysis with comparable historical patient data to provide information on the outcomes of different treatment approaches. Such detailed knowledge could help improve the precision of hip conservation procedures, allowing surgeons to avoid over or under corrections. “Our goal is to provide detailed information so surgeons can plan the best treatment for each patient,” says Kiapour.

A unique collaboration and a promising future

Kiapour points out that VirtualHip is not the first platform to analyze hip impingement, but it offers unique capabilities, including the automatic generation of detailed measurements without relying on a technician. It can also assess a patient’s risk of impingement or instability across a wide range of motion and process historical patient data to provide personalized analysis and treatment suggestions.

The insights underlying the development and refinement of these and other capabilities have emerged from an ongoing collaboration between Boston Children orthopedic surgeons, radiologists, and engineers. For five years, Kiapour and his team have worked closely with Dr. Kim and Dr. Edward Novais of the hip preservation team and Dr Sarah Bixby from Radiology.

We are blessed to work in an environment where multidisciplinary collaboration is strongly encouraged, and clinicians are eager to participate in translation research.”

“Clinicians are the end users of this platform, so keeping them updated throughout development is imperative to ensure the platform meets their needs,” says Kiapour. Working together, the team has developed a technically sound platform that provides clinically relevant information in a clinician-friendly format. “We are blessed to work in an environment where multidisciplinary collaboration is strongly encouraged, and clinicians are eager to participate in translation research.”

Must Read
Real-time rip current identification tool uses AI and deep learning

The combined team will continue to develop and validate the prototype so it can be used in a clinical setting. The ultimate goal is to make VirtualHip available for wide use in hip treatment centers. The project could also herald a new application of data science in pediatric orthopedic care: using AI to make lessons from historical patient data accessible to surgeons as they plan care for their patients.

Learn more about the Hip preservation program in children and young adults and the Musculoskeletal Informatics Group.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments