The important role of artificial intelligence (AI) in enterprise imaging workflows

Innovative technologies, like artificial intelligence (AI) and machine learning, are already helping health care systems lower costs and achieve better outcomes. Large volumes of image data and faster, more complex computer processing capabilities have made it possible to apply machine learning and pattern matching to medical imaging.

Right now, AI is currently being used both for efficient workflow management and imaging analysis. The two uses support one another, as AI-enabled image analysis supports better clinical insights, lower clinical variability and assist in studying prioritization. By focusing on higher-priority cases, radiology practices improve customer satisfaction and reduce patient wait times.

With improved data access and AI-driven worklist prioritization, care providers can better collaborate, improve labor utilization and increase daily productivity. 

When AI is used with a full enterprise imaging platform that is consistent across an organization, it can also help ensure the most value is gained from IT investment while helping to reduce overall costs. These innovative solutions provide the flexibility to meet evolving clinical needs.

With AI as a driver, provider organizations can determine three key benefits:

• Faster, better patient outcomes
• Automated study prioritization
• Measured and enhanced performance

Faster, better patient outcomes

For many time-sensitive conditions, like acute ischemic strokes, the time it takes to receive treatment is a key factor of clinical outcomes. CT scans must be done within 25 minutes of the patient’s arrival in the ER and the scan interpretation must be completed within 45 minutes, prior to beginning treatment.

AI technology can perform a thorough image analysis for specific clinical findings, such as analyzing CT head exams for anomalies related to intercranial hemorrhages. When radiologists are notified of an AI-detected acute abnormality, the care team can act quickly to treat the potentially life-threatening cases. 

Automated study prioritization

Prioritization is key in any profession, but it’s especially important in a fast-paced clinical setting. AI-enabled systems drive efficient worklist prioritization by continually communicating the results of image analyses. AI can help ensure that such studies are automatically assigned to the proper physician. 

Measured and enhanced performance

With one unified worklist, an AI-driven workflow intelligence solution can consolidate quality, communication and interpretation. This solution can help measure and improve productivity, drive accurate and efficient imaging and prove the overall value of the enterprise imaging department to the entire health system. 

A comprehensive workflow solution can provide a wide range of services including Radiology-ED communication, mammography imaging review, technologist QA and an anonymous peer review to help ensure that quality is always a top priority.

There are three fundamental steps to AI-enabled workflow prioritization. First, the system analyzes images to look for specific clinical findings. Then, it communicates the results by sending out alerts for high-priority studies that show abnormalities. Finally, it assigns a higher priority on the interpretation worklist to studies with noted abnormalities.

The system assigns priority to studies in the workflow based on patient setting, sub-specialty, procedure type, complexity, eligibility, service-level agreement (SLA), escalation, age and more. The most urgent tasks are displayed at the top of the worklist, so that users can prioritize the most critical tasks first. Instant messaging, notifications, email and text tools help improve collaboration across the enterprise. 

AI technology allows significant improvements in computer-aided detection (CAD). For example, in mammography, CAD image analysis identifies and presents areas of tissue density as image overlays in the radiologist’s image viewer. Added tools are being developed to identify more complex patterns that may mark the presence of specific findings such as intercranial hemorrhage, C-spine fracture, or pulmonary embolism. 

New learning-based AI prediction models are being trained to identify patterns that human clinicians may not see. These algorithms produce more precise results than existing risk models and are especially helpful when identifying predictive, pre-diagnosis discovery of disease.

The latest artificial intelligence tools can find and identify specific elements with great accuracy as they were trained to do. At the same time, radiologists are naturally much better than computers at seeing and interpreting the whole picture – and recommending the proper course of action. The best role for AI-enabled imaging technology today lies in the daily support that these tools can provide to radiologists.