Search
I once heard a chief radiologist of an academic institution speaking at an artificial intelligence roundtable describe his wish for AI to “be the best fellow he has ever had.” While I don’t think we are quite there yet, AI has been proven to provide many benefits to diagnostic imaging users, namely radiologists and cardiologists. We have seen a significant increase in the interest and adoption of imaging AI over the past year as the benefits of AI usage are realized. Those imaging professionals already actively using AI are sharing their experiences with other colleagues creating more interest in ever broadening circles.
One trend we have seen at Fujifilm is an expanded usage of AI in facilities that already are using imaging AI. As more clinical users adopt AI into their workflows and experience the benefits of increased diagnostic accuracy and more timely intervention through early notifications, they want to expand their usage into other diagnostic areas to take advantage of the benefits in those areas as well.
Now trending: AI Suites, Platforms, & Orchestrators With the increased interest in expanding imaging AI usage to more areas, another trend we have seen is the rising adoption of AI “suites” or AI “platforms” that can host multiple algorithms and vendors. An AI suite is typically a solution that was created by an AI algorithm vendor to host multiple algorithms scanning multiple areas of interest. For example, a suite could host algorithms from one or multiple vendors for the detection and prioritization of a variety of different disease states like intracranial hemorrhage (ICH), pulmonary embolism (PE), brain aneurysms, and bone fractures, moving suites toward an AI platform. AI platforms were traditionally created by a vendor that doesn’t manufacture their own detection algorithms, but rather pulls together algorithms from multiple AI vendors that again, can scan multiple areas of interest. In either case, the suite or platform provides a solution that allows healthcare providers to more easily obtain and deploy imaging AI algorithms, thus making it easier for them to expand usage into more areas.
Another trend we have seen is that workflow complexity has significantly increased with the rise of multiple AI algorithms being deployed, whether through a platform, suite, or directly. This is where using an AI workflow orchestrator, such as the Synapse® AI Orchestrator becomes invaluable. The Synapse AI Orchestrator can support and manage multiple scans conducted on the same study, even asynchronously. It supports various status states, error handling and multiple results being returned on these studies to keep the clinicians informed of the AI processing. Another recent change allows for modality-deployed or AI platform or suite workflows to be supported seamlessly within Synapse PACS in addition to the traditional PACS driven workflows. In either case, the results are incorporated directly in the PACS worklist and viewer.
AI Orchestrators in Action: Jefferson Einstein Hospital One example of the flexibility of the Synapse AI Orchestrator is our integration at Jefferson Einstein Hospital in Philadelphia. Jefferson Einstein Hospital has been a leading adopter of imaging AI in their radiology workflows and have integrated Synapse PACS with a wide number of imaging algorithms. Jefferson Einstein Hospital recently partnered with Fujifilm to deploy the Synapse AI Orchestrator in their workflow to tighten the integration with the AI algorithms and bring the results natively in the PACS with additional worklist status, priority, and preview features.
The first integration with Synapse AI Orchestrator was with Aidoc, an AI suite provider. Jefferson Einstein Hospital integrated six individual detection algorithms with Synapse AI Orchestrator to begin with PE, ICH cervical spine fracture, rib fracture, abdomen free air, and brain aneurysm. The radiologists report that having a worklist indicator of a significant finding from the AI scans has helped improve patient treatment. In several reported instances a critical finding indicator on the worklist led the radiologist to review the case sooner than they would have otherwise. In multiple cases, they were able to refer to the patient for immediate treatment for an acute condition much sooner than would have otherwise occurred. Based on the success of the integration with the initial six algorithms, Jefferson Einstein Hospital plans to expand integration with additional algorithms through Aidoc, Gleamer for fracture detection, and Riverain for CT lung nodule detection and vessel subtraction.
As AI in medical imaging continues to adapt and improve over time, the adoption of AI will continue to advance as well. The overall value of AI in medical imaging extends beyond diagnostic and treatment benefits. It plays a crucial role in advancing medical research by providing tools for the analysis of large datasets and identifying patterns that may be challenging for researchers to discern. As AI contributes to ongoing development of innovative therapies and interventions, Fujifilm’s AI strategies will continue to stay abreast of new medical discoveries and evolving healthcare challenges and be ready to meet them head on.
Get hands-on with Fujifilm’s AI Orchestrator at the upcoming Radiological Society of North America (RSNA) conference held November 26 – 30 in McCormick Place in Chicago. Be sure to plan a visit to the Fujifilm booth #1929 – you can book your demonstration here. I look forward to seeing you there!