An imaging suite/department has many moving parts. There are several steps involved:
- patient scheduling
- protocol selection and patient preparation
- tracer/agent administration
- image acquisition & processing
- image interpretation & reporting
- radiologist performance analytics
Let’s review these to see how AI can play a role therein.
There are many software solutions on the market today that have attempted to streamline and automate the process. The main goal is to reduce downtime and optimally utilize resources, and in addition, ensure an efficient and reliable way for patients to make their preferred selections. AI can play a role in helping the patient find the most desirable time slot while ensuring that the resources such as camera time and personnel are optimally utilized. Furthermore, it can send automated follow-ups and help in rescheduling for new time slots as needed.
Protocol selection and patient preparation:
There is a wide variety of protocols used in adult and pediatric imaging. Choosing between them can be hard and several factors come into play. There are instances where wrong protocol selection leads to suboptimal images, overdoses, patient discomfort or even morbidities. AI can (semi-)automate the process by which the system can use the available information, such as patient age, sex, body habitus, serum levels, allergic/medication history, prior imaging, prostheses, other contraindications etc. and provide the most suitable protocol. This can ensure patient safety and desirable imaging results leading to superior quality of medical care.
Image acquisition & processing
AI is playing a big role here. There is a lot of research showing improved image acquisition using AI, whether it is MRI or PET. Improved acquisition techniques have led to faster imaging time, lower radiation dose, higher resolution, etc. Post-acquisition processing has benefited from AI as well, including 3D-iterative techniques and kinetic modeling, etc. AI continues to grow by leaps and bounds in this space.
AI has the potential to impact radiology information system (RIS) and picture archiving and communication system (PACS). AI-based hierarchical worklists, nuanced feedbacks, trend analytics, lesion tracking, synoptic processes are in the pipeline and once implemented can greatly improve the clinical radiology workflow.
Image interpretation & reporting
The role of AI is quite prominent here as well. From CAD to radionics, AI is innovating automated lesion analysis to unravel subvisual information on disease processes. Likewise. AI-driven semi-automated structured reporting that checks for errors in real-time and provides instant feedback are poised to streamline the reporting process and make it safer, accurate and informative. AI can also aid in the radiologist-referrer communication loop in order to ensure effective transfer of critical knowledge to the ordering doctor for immediate action as needed. This reduces the chances of miscommunication that may lead to clinical mismanagement.
Radiologist performance analytics:
Finally, AI is able to significantly improve the existing radiology performance system by introducing novel but useful metrics, such as hedging scores, feedback scores, biopsy results, protocoling errors, etc., and help create a more holistic performance evaluation of the radiologist, which can accurately identify deficient areas in clinical practice.