Evolution of Radiology
According to recent studies,with the Evolution of Radiology the radiologists can improve accuracy & speed of elucidating diagnostic images like CT Scan and X-rays.
Though certain complexities come into picture for example, testing, requiring clearance from regulatory authorities and development. But, like ‘the app store’ there is a concept of Radiology AI Marketplace which is very much similar to the former in distribution, enabling discovery, monetization of AI models (in former case ‘Apps’) & creates a better bridge of feedback channel between developers and users with the Evolution of Radiology. These marketplaces differ from the app store in supporting the life-cycle requirements for taking regulatory approval, training, developing & validating AI models.
If we talk about the practical use of AI in healthcare organizations, Radiology algorithms deal with one scenario at a time or in other words it focuses on single finding from imaging modality for example, lung nodules on a chest CT and this is an advantage for dealing with specific cases and requires purpose-built algorithm. So, developers need to generate, train, test, support, distribute, update algos and seek FDA approval. And healthcare firms need to find, evaluate, procure and deploy the algos in workflows. AI models are highly data driven and algos are as robust as data in which they are developed. So, if an AI model is working well with one data environment then it may lose its strength when used with other locations with different populations, techniques & imaging machines.
AI marketplace takes care of these challenges like by providing hospital systems one-stop access to many AI models. Also, there is a built-in feedback channel through which Radiologists can share the results with the developers so that to update the algos in real time using annotated and data. This also helps in getting validation data for FDA approvals. In addition, hospital systems also get the access to track algo usage, costs and performance from multiple locations. University of Rochester uses an application developed by Aidoc. This application prioritizes the CT exams (for suspected intracranial hemorrhage) whenever the time to treatment is critical. Likewise, University of Pennsylvania is using an application developed by eUnity & Aidence to assist radiologists in detecting & characterizing lung nodules which can be a time-consuming process without this AI model. FDA’s Software Pre certification Program (SPA) is working with developers to provide them efficient ways to develop new AI models. Another initiative of FDA is National Evaluation System for health Technology (NEST) is working with the American College of Radiology to develop ways to monitor and validate algorithms’ performance for detecting & classifying lung nodules in lung cancer screening programs.
This way AI marketplaces can provide efficiency and accuracy in the treatments and help in managing workloads. As the future will evolve, the need for advanced technologies will surely increase. AI is just another milestone.
Futurios technologies can help you in implementation or artificial intelligence for your Healthcare business, Get in touch for a consultation at [email protected]
Written By Ajay Sharma Strategy Consultant Futurios technologies