The global computer vision industry is expected to increase fourfold over the next 10 years, seeing a growth in value from $9.45bn in 2020 to $41.11bn by 2030, according to Allied Market Research.
Advances in technology have made computer vision competitive with human vision, in terms of pattern and object recognition, across many domains — from self-driving cars to healthcare. Computer vision is on track to transform the healthcare industry through its application in imaging diagnostics, post-surgery tracking and patient symptom tracking.
Investment in computer vision projects in healthcare is growing on par with the popularity of the technology. The market for computer vision technology in healthcare shows an annual growth rate of 47.2% and is projected to reach $1.46bn by 2023, according to estimates from Markets and Markets Research.
In this article, you’ll learn about some of the most promising applications of computer vision in healthcare, and their benefits for the healthcare industry’s growth.
But first, let me give you quick definitions and a brief overview of the key market players.
Have you ever used face recognition to unlock your smartphone?
If so, you’re already familiar with computer vision. Face recognition — probably the most popular use case of computer vision — identifies a person using a smartphone camera as human eyes to detect who it sees. This is similar to what human vision does when we encounter a face, identify an object or scan the surrounding environment.
Basically, computer vision is a form of AI that uses algorithms to replicate human visual recognition abilities. Algorithms are trained to analyze images and recognize patterns in them with the goal of identifying and classifying objects.
The most common applications of computer vision in healthcare are related to medical imaging, helping doctors detect diseases and pathologies from X-ray, CT and MRI scans.
As mentioned above, analysis of the market for computer vision in healthcare carried out by Markets and Markets Research forecasts an increase in value from $210m in 2018 to $1.46bn by 2023. They forecast steady growth across regions, with North America at the forefront.
The most prominent players in this domain are IBM, Intel, NVIDIA, Google, Microsoft, Xilinx and iCAD. While most of these big names are associated with disruptive technologies other than healthcare, they have been actively contributing to computer vision projects in healthcare for some time.
For example, NVIDIA, famous for producing graphics processing units (GPUs), partnered with King’s College London, one of Europe’s leaders in medical research, on several projects related to computer vision in healthcare. One of those projects is an AI platform built on the NVIDIA DGX-2 AI system to help UK NHS specialists automate some parts of radiology interpretation.
Intel collaborated with the pharmaceutical company Novartis to use deep neural networks for image analysis in drug discovery.
Google went further, creating a healthcare subdivision within the corporation. One of their projects focused on developing a tool that uses a deep learning system to examine images of the eye to detect evidence of diabetic retinopathy. Worldwide, diabetic retinopathy is responsible for most cases of vision loss. However, despite showing high theoretical accuracy, the tool failed in real-life testing.
Now, let’s take a closer look at how computer vision can transform healthcare through its application in radiological diagnostics, medical imaging and post-surgery blood-loss tracking.
The ability of computer vision to detect patterns in images, including X-ray, CT and MRI images, has potential to help doctors make more accurate and timely diagnosis of various diseases.
Deep-learning-enabled solutions in this domain rely on neural network models trained to classify medical images. They can detect lesions, tumors and suspicious patches that require further examination.
In particular, substantial progress in this domain has been made related to detecting lung cancer. In an article in Nature Medicine, a group of scientists from Google AI, NYU, Stanford, and Northwestern reported that their deep learning model trained to detect lung cancer outperformed trained radiologists, in terms of detecting both false negatives and false positives. Moreover, using prior CT scans, their tool could predict the risk that a patient would develop cancer, based on the growth of nodules over several years. The system is one of several moving toward clinical adoption now, according to Nature.
The hope is that computer vision will help health professionals detect other forms of cancer with higher accuracy. Several independent research studies on the application of neural networks in melanoma detection indicate that computer vision models can diagnose on par with doctors’ accuracy.
The Pneumonia Detection Web App is another example of a neural network. It analyzes chest X-rays to determine if a patient has pneumonia. According to developers, the model showed 86% accuracy on a test set of over 500 X-Rays. However, it hasn’t been clinically validated and is used for research and comparison purposes only at this point. The code and other details are available in an open source repository.
Thank you for reaching out,
User!
Make sure to check for details.
Medical imaging (non-radiological) is another domain where computer vision can contribute to early detection of disease. In this case, visualizations allow doctors to look at a patient’s tissues and organs more comprehensively to identify abnormalities.
The Arterys medical imaging AI platform, for example, offers hospitals and medical diagnostic imaging centers software to help detect abnormalities in the heart, lungs, and brain. It displays 3D models of a patient’s organs on a radiologist’s computer screen. Its algorithm helps detect patterns in the 3D models that can lead to diagnosis of conditions at their early stages.
Computer vision also has the potential to reduce post-surgery mortality rates, by tracking patient blood loss in real time, for example.
A California-based med-tech company, Gauss Surgical, developed Triton, an AI-powered monitoring solution that helps doctors accurately estimate blood loss after a woman gives birth, including C-section births. The system captures images of blood-containing canisters and sponges and applies computer vision to estimate the level of blood loss with higher precision than physicians can, according to the developers.
Usually, doctors estimate blood loss visually, which is often inaccurate. Inaccurate estimates can lead to aggravation of a woman’s condition, and even death. The use of computer-vision-enabled systems to track blood loss can reduce maternal mortality rates caused by undetected hemorrhage.
Triton has been approved by the US Food and Drug Administration (FDA) and Gauss Surgical boasts numerous clients on their website, including University of Texas Medical Branch, Hackensack UMC and UC Irvine.
Mount Sinai Hospital in New York, reported a 340% increase in hemorrhage detection since they’ve been using Triton, according to Gauss.
Technological advances and scientific research on potential applications of computer vision in healthcare show that the technology is set to transform the industry, yielding numerous benefits for medical institutions and patients.
Early detection of diseases and pathological conditions, highly accurate diagnosis, reduced patient examination times, and the prospect of home-based diagnostics and monitoring are some of the advantages of integrating computer vision into the workflow of medical institutions.
It’s worth mentioning that no technology can replace medical professionals. However, technology can serve as a valid tool to assist in streamlining workflows at medical facilities, increasing the accuracy of diagnosis, and improving treatment quality for patients.
If you’re looking for a technology partner for your computer vision projects in healthcare, don’t hesitate to contact us. Our machine learning engineers will help you bring your ideas to life.
