AI & Optical Coherence Tomography

We’ve seen growing advances in AI for space and defence, automotive, and consumer electronics, but where does Medical Devices currently stand?

While the industry is still navigating key security regulations, especially for HIPPA compliant cloud systems, AI solutions for optical coherence tomography have proven to be the stand-out player.  

According to Absolute Market Insights, the global AI radiology market was valued at $187.61 million in 2018 and is anticipated to reach $3506.55 million by 2027. CT and MRI imaging are the first to take advantage of AI techniques. Deep learning, particularly convolutional neural networks, have been the popular choice for optimizing radiology. By layering images, extracting features, and aggregating that data, we are now able to completely segment organ systems and quickly identify lesions and pathology. Thus, allowing extremely early detection, optimized work-flow, and productivity. Besides using machine learning to reduce noise and create high-quality images with low radiation, it also has the potential to reduce tedious tasks for radiologists, offer a second opinion in terms of diagnosis especially for intra-observer variability and inter-observer variability.  

So, who are the major players?

Arterys, located in San Francisco, built the first tech product to visualize & quantify blood flow in the body using any MRI. Arterys also received the first FDA approval for clinical cloud-based deep learning in healthcare. Furthermore, they’ve pioneered four-dimensional (4D) cloud-based imaging. 

GE Healthcare and NVIDIA announced collaboration in 2017 for a new CT system (FDA approved) that is “two times faster” than the previous system and is anticipated to more quickly to identify liver and kidney lesions due to the high volume of data accessible through NVIDIA’s AI platform. Increased speeds mean less radiation and less time which can equal faster treatment and better clinical outcomes overall.

What’s next? 

Greater development in unsupervised learning for rare diseases. It’s nearly impossible to detect rare diseases using newer AI programming. Developers and researchers are focused on solving unparalleled segmentation power and greater classification. Currently, the FDA has approved 28 AI algorithms for tomography, including approved algorithms for several medical device companies Meet has had the privilege of partnering with. Given the demand and incredible potential, Meet will continue keeping a keen eye on who will be emerging leaders for AI and tomography.