Imaging devices play an important role in diagnosing diseases, yet the crucial step of image interpretation and disease diagnosis requires a human expert. This age-old setup has imaging devices outputting an image and human experts taking the images and outputting diagnoses.
But all that is changing with AI.
A medical imaging device is a specialized equipment used to visualize the interior of the human body for diagnostic purposes. Imaging devices produce images of organs, bones, soft tissues, and blood vessels which helps physicians in diagnosing and treating diseases. There are many different types of medical imaging devices, each with its own advantages and limitations. Some of the more common types of medical imaging devices include X-rays that use ionizing radiation to produce images of bones and dense tissues; MRIs that use strong magnetic fields and radio waves to produce detailed images of organs, soft tissues, and bones; ultrasounds that use sound waves to create images of organs and soft tissues; and retinal cameras, also called fundus cameras, that are specialized photo cameras used by eye specialists to capture high-resolution digital images of the interior of the eye, specifically the retina, optic nerve, and macula.
As of today (2024), there are numerous AI solutions that integrate with imaging devices to assist human experts in the task of interpreting the images and diagnosing diseases from them. For example, AI technology is built into the post-processing of the images to segment areas of interest for the expert to pay attention to, or flag images determined by the AI to require immediate attention because of a life-threatening condition.
But for at least one medical domain - diabetic retinopathy screening using retinal cameras - AI has been cleared by the FDA to autonomously diagnose the disease instead of a human expert. To-date, the FDA cleared three AI solutions to diagnose diabetic retinopathy from retinal images.
Once an AI with autonomous diagnostic capabilities is integrated into a retinal camera, it transforms the camera from an ordinary imaging device to a diagnostic device. Instead of the camera just producing images as its output and sending the images for interpretation by an expert, whose output is a diagnosis, the AI-enabled retinal camera outputs a diagnosis - a final diagnosis that does not require a human expert to review or sign off on. An exceptional example for such a solution is AEYE Health’s autonomous AI that is integrated into Optomed’s handheld retinal camera - take a look:
Moving forward, more AI solutions that diagnose instead of an expert will receive the green light from regulators, paving the way for more imaging devices to transform into diagnostic devices.
Comments