FUJIFILM SonoSite, Inc. announced the launch of a strategic relationship with Partners HealthCare to apply artificial intelligence (AI) to POCUS applications. The two companies will collaborate to use AI-based models that will assist clinicians in diagnosis and enhance quality of care. The collaboration will be executed through the MGH & BWH Center for Clinical Data Science and leverage the extensive data assets, computational infrastructure, and clinical expertise of the Partners HealthCare system.
Rich Fabian, President and Chief Operating Officer of FUJIFILM SonoSite, Inc. said of the new venture, “I am proud that we will be able to offer our customers AI-enhanced technology to expand their utilization of ultrasound, increasing the quality of care they can provide while saving our healthcare system money.”
Dr. Diku Mandavia, Senior Vice President and Chief Medical Officer of FUJIFILM SonoSite, Inc. looks forward to the innovations fostered by the partnership.
“This collaboration is really focused on embedding AI in portable ultrasound with the goal of providing assistance in 2D image interpretation along with the type of automation that will allow us to increase the accessibility of this critical technology while still delivering high diagnostic value."
In a May 2019 article posted by the POCUS Certification Academy, Dr. Victor Rao expressed his vision of the merger of AI with POCUS, which touched upon the advances Dr. Mandavia mentioned. Dr. Rao begins by addressing the areas of experience and training. To evolve into a POCUS expert, you must have a “practice makes perfect” mindset, as Dr. Rao explains:
“Ultrasound is a hands-on modality, which means you have to scan frequently to become more efficient at scanning patients, as well as scan difficult or challenging patients whose anatomy is not as easy to visualize with ultrasound due to body habitus or some deformity.”
This is where AI technology comes in. AI can assist newer users of POCUS by guiding them in acquiring the right kind of images. Learners can take advantage of AI features to help them recognize life-threatening conditions, assist with probe alignment, and obtain the next steps in patient management. Dr. Rao believes that this assistance will make POCUS users eager to learn more. AI technology can also make the POCUS user’s analysis far more efficient by rapidly comparing thousands of scans from past patient records. Dr. Rao describes this innovation:
“You don’t have to have that much experience to be able to tell what the ejection fraction of the patient’s heart is. The AI component can compare your patient’s images and video loops with thousands or even hundreds of thousands of other patients’ data to assist you in drawing an opinion.”
Dr. Rao sees ultrasound image optimization as another area where AI has an important role to play. AI technology can ensure the right levels of image brightness or darkness, as well as adjust depth control or prompt the user that the depth needs to be corrected. Optimal images are essential in making the correct diagnosis and not missing any crucial information.
But even with all these machine-led improvements, the joining of AI with POCUS will not eliminate the need for medical professionals. For example, some conditions are atypical, so there may not be sufficient data for AI to identify a rare condition. A clinician with a knowledge of anatomy, pathology, and physiology will always be needed to perform POCUS procedures, and AI relies on human ultrasound users to generate the data it utilizes. And, of course, AI cannot replace the human connection that all patients need. Patients will always rely on and expect empathy and understanding from their health care providers.
Physicians should aim to adopt AI technology to provide the best possible care to their patients. “Allowing for even greater integration of ultrasound into our healthcare delivery system requires smarter machines,” says Keith Dreyer, Chief Data Science Officer of Partners HealthCare. “In emergency settings, the efficiency and cost-effectiveness of portable ultrasound make it preferable to other imaging modalities.”