This is the initial frontier of AI in healthcare. Why? images are for the most part digital files with structured data that can be used to develop and validate a model to perform a narrow task such as finding tumor on a CT scan or a fracture on an X-ray. The infrastructure that has been created for digital imaging yields itself well to incorporating algorithms into that workflow. That means it is possible, not easy!! Health systems have many priorities with limited resources so incorporating an AI algorithm into the radiology systems is still not easy for them. However, out of the many possible applications of AI in healthcare, this is one that is currently possible. In addition to radiology, some of the other early applications of AI in healthcare includes the use of images to screen for or diagnose conditions based on images. Not necessarily radiology images but images from a cell phone, such as facial images for stress levels or congenital conditions or skin images for dermatologic conditions.
Medical imaging is the largest and fastest-growing data source in the healthcare industry. it accounts for 90 per cent of all healthcare data. What is even more astonishing is that more than 97 per cent of it goes unanalyzed or unused. It is estimated that there were 4.7B diagnostic imaging scans performed globally in 2019 (including in- house, using radiology groups and using teleradiology reading service providers), a number that is forecast to grow at approximately 3% per annum over the next five years. X-ray exams accounted for most of these scans in 2020, but others such as MRI and CT are forecast to grow at a faster rate over the coming years, as they have over the last decade.
Reading these scans involves radiologists carefully studying each study and finding abnormalities based on their training. There are 800 million medical scans in a year that result in 60 billion images. There is estimated error rates of 2% false positive and 25% false negative in reading these scans. Processing this massive volume of medical imaging data could lead to longer turnaround times from image acquisition to diagnosis to care. Meanwhile, patients’ health could decline while they wait for diagnosis. Especially when it comes to critical conditions, rapid analysis and escalation is essential to accelerate treatment.