Making health care more equitable one ultrasound image at a time

AI-powered portable ultrasound system makes scans and diagnoses accessible from anywhere.

220428-ai-ultrasound-td-jacob-jaremko-main-16x9-2000px.jpg

Jacob Jaremko is leading a project to expand access to a new, AI-powered portable ultrasound system he developed, thanks to funding from TD Bank Group. (Photo: Laughing Dog)

Imagine if family doctors, nurses, technicians — even ski patrollers — could use handheld ultrasound devices to screen for everything from hip dysplasia to wrist fractures to heart disease from anywhere.

A new approach to scans developed at the University of Alberta pairs the accessibility of portable ultrasound with an artificial intelligence app that can analyze thousands of previous results and provide remote diagnostic support within seconds.

The research will be expanded and the system tested across Alberta thanks to a new $450,000 grant from TD Bank Group (TD) announced today. It expands on a pilot study funded by Alberta Innovates, which demonstrated the feasibility of using the technology in primary care.

The goal of the project is to make access to health care more equitable and improve outcomes for patients, said project lead Jacob Jaremko, a radiologist, associate professor in the Faculty of Medicine & Dentistry and Canada CIFAR AI Chair at the Alberta Machine Intelligence Institute (Amii).

“It really is a whole different model of care,” Jaremko said. “You're taking expertise learned from the experts in the hospital and delivering it to the patients, rather than having the patients come to the hospital.”

“The work being done at the University of Alberta to advance portable ultrasound capabilities and pair it with AI technologies will help improve access to timely medical diagnoses and care, particularly for those living in remote and Indigenous communities,” said Robert Ghazal, TD senior vice-president for branch banking in the Prairie region.

“Through the TD Ready Commitment, the Bank's global corporate citizenship platform, we are proud to support innovative solutions like this one that will help create more equitable health outcomes for all.”

Saving time and money

Ultrasound is one of the most commonly used diagnostic tools in Canada, but scans are usually carried out in major hospitals or specialized clinics, and wait times can range from weeks to months because of a shortage of specially trained sonographers and radiologists. Studies have shown that patients in low-income neighbourhoods and rural, remote and Indigenous communities access ultrasound less than other Canadians.

With Jaremko’s system, a non-expert can learn with brief training how to use a portable ultrasound device to scan for thyroid cancer, hip dysplasia in infants, breast cancer, heart and lung problems, and arm fractures. The images are uploaded and the AI app compares them with the results of many previous tests to determine whether there is an abnormality.

“The artificial intelligence network trains using hundreds or thousands of cases. It sees those cases through the eyes of the human experts who have read all those cases, and it takes the distilled lessons of that human expertise to provide diagnostic support,” explained Jaremko, who co-founded MEDO.ai to commercialize the technology.

Jacob Jaremko explains how his AI-based ultrasound system works and why it could increase access to vital diagnostic support for conditions like hip dysplasia. (Video: MEDO.ai)

This can save a lot of time for the patient and money for the health-care system,” said Julie Hernberger, a family doctor in Red Deer who has scanned about 40 newborns’ hips since October using the handheld ultrasound device plugged into a tablet.

Hernberger takes less than a minute to do the scan, uploads the images and gets a result back immediately. In the past, any babies in Red Deer suspected of having hip dysplasia had to be sent to Edmonton or Calgary for formal ultrasound scanning.

Family doctor Julie Hernberger demonstrates a hip ultrasound procedure on her four-month old granddaughter, Olivia Dyck. (Photo: Jessica Kaitlyn Photography)
Family doctor Julie Hernberger (right) demonstrates the hip ultrasound procedure on her four-month old granddaughter, Olivia Dyck. (Photo: Jessica Kaitlyn Photography)

Hip dysplasia occurs in about one in a thousand births and is more common in females, breech babies and some ethnic groups. It is easily treatable if caught early but can lead to lifelong arthritis if it isn’t. Hernberger has one young adult patient whose dysplasia was not diagnosed until her late teens and now faces complex bilateral femoral osteotomies and eventual total hip replacements.

“I'm hoping that if I can scan every baby I take care of, if I can catch even one of these babies and prevent them from having to face that kind of surgery when they're 20 years old, then I feel like I've accomplished something great,” said Hernberger.

Family doctor Julie Herberger holds her four-month-old granddaugher, Olivia Dyck. (Photo: Jessica Kaitlyn Photography)
Julie Hernberger with her baby granddaughter, Olivia Dyck. Hernberger hopes to use the AI ultrasound system to scan her young patients for hip dysplasia, which is easy to treat if caught early but can lead to lifelong arthritis if it isn't. (Photo: Jessica Kaitlyn Photography)

More accessible, equitable care

The TD grant will allow Jaremko’s team to gather more comparator data, improve the algorithms and implement the system at clinics across the province. So far the hip dysplasia ultrasound system is being tested at Westview Primary Care Network in Spruce Grove and Saint Mary Clinic in Red Deer, but Jaremko is looking for other clinics, particularly in northern Alberta towns and Indigenous communities, to test the system.

“The further away you are from the main hospital — either geographically or socio-economically or culturally — the more likely this will be to help you,” he said. “It really enhances the equity of care for Albertans.”

Jaremko’s lab includes several graduate and undergraduate students from both computer science and medicine, and he expects to recruit more thanks to the new funding.

“We are an ideas lab, and students really like to work with us because it's bench to bedside — you can work on an AI algorithm and then you can try scanning a patient with it,” he said.

The team’s system to scan for thyroid cancer has already been approved by the FDA in the United States, and the hip dysplasia system is under consideration for approval from Health Canada. A basic echocardiogram system that looks at heart function is also in development.

“It'll never replace the complete echocardiogram, but it will be a screening tool that says either things look pretty normal, or else things are not normal and you need to come into hospital to get this checked.”

Jaremko’s team is also developing a lung scanning system that could detect pneumonia, excess fluid or even COVID damage, and their system to identify wrist and elbow fractures is being tested by ski patrollers at Alberta hills. Jaremko is also working to ensure scan data remains accessible to researchers while maintaining patients’ privacy.

“We are trying to get AI out of an academic sphere, or even out of a tertiary hospital, and into real use in the community,” he said. “Patients will experience a direct improvement in their health as a result.”

Jacob Jaremko held the Alberta Health Services Endowed Chair in Diagnostic Imaging at the U of A from 2011-2021 and is a partner at Medical Imaging Consultants. His research is also supported by the Stollery Children's Hospital Foundation through the Women and Children's Health Research Institute and Alberta Innovates.