Whilst X-rays accounted for most of the scans performed, it is estimated that they accounted for less than 20% of total radiologist reading time, due to faster reading times per scan. The net effect of the changing complexion of scan types forecast over the next five years is, not only are diagnostic procedures increasing, but procedures that take a disproportionately longer time to report are growing fastest, increasing the demand for radiologist resource. AI offers a huge competitive advantage for teleradiology reading service providers that can reduce these read times, whilst maintaining (or even better, improving) accuracy.
Teleradiology reading service providers have been exploring how AI can be used to improve workflows and decision support. For example, US teleradiology service provider, vRad, has also been working with Qure.ai on worklist prioritization, specifically in relation to intracranial hemorrhage on head CT scans. Other examples include Real Time Medical of Canada collaborating with Google Cloud to develop AI-assisted workload balancing tools; I-Med Radiology Network of Australia implementing AI tools for worklist triage in the areas of brain hemorrhage, pulmonary embolism and C-spine fracture; and Global Diagnostics Group, again of Australia, partnering with Aidoc, to implement AI-based workflow solutions supporting care management pathway development.
The potential benefits of AI in radiology are not limited to better and faster reading of the scans. AI can actually help acquire better images or high-quality images faster. Facebook has an MRI initiative that has developed models that can generate equally accurate and detailed MRIs using about a quarter of the raw data traditionally required for a full MRI. Since less data is required, MRI scans can run nearly 4x faster. A team of independent radiologists compared the AI-generated images with traditionally captured images and could not tell which were created using the new method. Subtle Medical has developed AI technology that use image processing techniques to speed up MRIs and PET scanning processes and enhance the quality of images obtained.
Also, a number of companies are embedding AI in ultrasound machines that help primary care offices to perform the study without having received full training. The AI can guide the user in acquiring better quality images. Examples of these companies include butterfly, caption, Exo.
