The top four advantages of AI in cardiovascular ultrasound

Just as in the non-medical world, the key to successfully harnessing AI in ultrasound is knowing where to focus it. When considering what is needed to make diagnosis and treatment easier, more effective, or faster, Philips employs a clinician’s perspective. This perspective results in AI ultrasound that is designed to integrate into every day clinical workflows, augment user skills and help reduce cognitive burden. It transforms AI into a valuable assistant – one that delivers outsized advantages in four areas: reduced variability, accurate measurements, decision support and simplified workflows.

Ultrasound is arguably the most user-dependent imaging modality. Small changes in angle or pressure can result in significant changes in diagnostic output. Well-chosen views can result in diagnosis, while poorly chosen ones may add time and cost by resulting in repeat exams or referral to other imaging modalities. New ultrasound AI applications reduce variability scan-to-scan and user-to-user by helping to eliminate the need for manual view selection and visual analysis. For example, SmartView Select runs behind the scenes while a clinician is capturing images, identifying effective images to use in applications including Strain or EF.

Echocardiography measurements are another area that benefits from AI-assisted accuracy. Two examples that illustrate the power of AI to perform clinically useful measurements are:

AI-driven automated measurements also reduce the number of keystrokes and clicks, [2,3,4,5,6] potentially alleviating the physical burden on clinical sonographers.

These consistent, automated measurements contribute to a third advantage of AI ultrasound: decision support. Today, clinicians can take advantage of technology that uses cardiovascular ultrasound AI to turn images and measurement data into analysis that supplements clinician expertise and helps increase diagnostic confidence. One excellent example of this is 3D Auto Color Flow Quantification, the first fully automated 3D quantification of mitral regurgitation (MR) volumes. Designed to provide reproducible, efficient analysis, it helps clinicians make better-informed decisions for patients with heart valve disease. A second example is Segmental Wall Motion Scoring, which calculates a score that aids assessment of LV impairment.

Automating routine tasks via AI supports a patient-focused care environment, frees up time for more meaningful tasks, and has the potential to help reduce the “rat race” mentality that has crept into imaging departments and serves no one – patients or clinicians. At an AI-focused RSNA 2024 plenary session, Nina Kottler, MD, MS, pointed out, “Our current processes and technologies just aren’t serving us. There’s a massive amount of information coming into the system, and our turnaround times are increasing” [7].

AI-fueled automation helps removes redundant tasks and helps saves time without harming outcomes. For example, Auto Measure, when used with the X5-1 transducer, reduced quantification time by a remarkable 51% [2]. Advances to AutoStrain feature fast, reproducible results as part of a comprehensive LV assessment within the same application, improving workflow and saving time.

Task automation also has the potential to ease the burden of staff shortages. Industry leaders are keen on these possibilities: 84% of industry leaders believe that automation will save time on day-to-day tasks, while 76% expect automation will enable staff to perform at their peak [8].

The four top advantages of cardiovascular ultrasound AI – reduced variability that decreases the user-dependent nature of ultrasound, consistently accurate measurements that ease diagnosis, decision support that adds peace of mind and simplified workflow through reduced repetitive tasks – are not promises of yet-to-be-developed technology. Each of these advantages is available today, through the 26 AI applications available on Philips cardiovascular ultrasound systems. To learn more, contact your Philips representative.