Computer Vision and the Evolution of Medical Imaging

To celebrate World Health Day we thought we would highlight some of the awesome ways that computer vision and the healthcare industry have joined forces to empower industry professionals. As the world population continues to grow and average life expectancy continues to rise, so too will the demand for healthcare workers. It’s been estimated that there will be a worldwide shortage of 15 million healthcare workers by 2030. Computer vision can offer viable solutions for lightening the human workload in the medical field. 

Computer vision and machine learning applications for the medical field continue to uncover the need for reliable data annotation tools. Medical imaging is the perfect example of why data annotation platforms need to be accessible to subject matter experts in order to produce the most accurate, helpful, and timely computer vision solutions–something we have kept in mind as we have built out the Sense Platform.

Caption Health has had great success in integrating computer vision and deep learning for cardiac ultrasounds. Their Caption AI ultrasound platform had a 92.5% diagnostic match rate when compared to ultrasounds taken by skilled cardiac sonographers. The platform has even obtained FDA clearance for cardiac ultrasounds at the point of care. The software is not meant to replace trained cardiac sonographers but to expand the pool of medical professionals who are able to perform the procedure. We could see this being useful in both over-burdened urban centers as well as in rural and remote areas where cardiac specialists are harder to come by.

An example of the Caption Guidance Technology. Courtesy of Caption AI’s website.

Another fantastic use case for computer vision is for monitoring of blood loss and hemorrhage prevention during childbirth. Historically, blood loss has been monitored through human observation–which can be inaccurate. Gauss Surgical’s Triton AI is able to examine surgical sponges and towels via an iPhone to calculate how much blood a patient has lost during labor in order to alert medical staff to the possibility of maternal hemorrhaging. The company reported a 2x-4x percent increase in hemorrhage detection when Triton AI was used. Postpartum hemorrhaging is the number one cause of maternal mortality worldwide.

Triton AI’s computer vision software can help medical staff monitor blood loss during childbirth. Images courtesy of Gauss Surgical’s website. 

Cancer diagnosis is yet another area where computer vision makes an excellent tool. Most cancer is ultimately diagnosed by examining stained cell biopsies under a microscope which requires the eyes of a skilled pathologist to identify the harmful cells; but computer vision and machine learning can help. PathAI has been able to successfully deploy machine learning algorithms to predict tumor molecular phenotypes across 5 tumor types. An incredible feat that the company says required over 1.6 million annotations of cancer cells

Because image processing is already such a huge part of modern medical diagnostics and aftercare computer vision driven solutions seem to be a natural evolution for the industry. We are excited to be a part of it. 

More Sixgill Blog Posts: