Caption Health Presents New Research on Caption Guidance Performance at ESC Congress 2020
Descriptive subgroup analysis finds Caption Guidance AI technology enables nurses without prior ultrasound experience to obtain diagnostic-quality images in patients with implanted electrophysiological devices.
BRISBANE, Calif., Aug. 29, 2020 /PRNewswire/ -- Caption Health, a leading medical artificial intelligence (AI) company, and its collaborators today announced additional research analyses from its landmark pivotal study indicating that its Caption Guidance™ AI-guided imaging technology can effectively guide novice ultrasound users in acquiring diagnostic-quality exams in patients with electrophysiological (EP) implanted devices. As Caption Health's AI technology expands access to high-quality medical imaging in a variety of clinical settings, including rural areas and primary care, this research is encouraging for an estimated 500,000 to 3 million patients in the US alone with an implantable pacemaker.
Caption Guidance provides real-time guidance and automated quality assessment to help users optimize and capture a diagnostic-quality ultrasound image, and was designed to emulate the expertise of a sonographer. The deep learning model used in the software was trained on over 5,000,000 observations of the impact of probe motion on image orientation/quality.
This research, which was presented at ESC Congress 2020, the annual meeting of the European Society of Cardiology, and was selected as one of the "Best ePosters", was a descriptive subgroup analysis from the pivotal study conducted to support Caption Health's landmark FDA De Novo authorization in February. In the larger study, eight nurses without prior ultrasound experience each acquired 10 standard transthoracic echocardiography (TTE) views in 30 patients (for a total of 240 total patients) guided by the novel Caption Guidance software. A pacemaker or implantable cardioverter defibrillator (ICD) was present in 27 of these patients. On the same day, trained cardiac sonographers obtained the same 10 TTE views without using the deep learning algorithm. A panel of five Level 3 echocardiographers independently assessed the diagnostic quality of the exams acquired by the nurses and sonographers.
The descriptive analysis evaluated performance of the software for patients with implanted EP devices, focusing on right ventricular (RV) size and function. Nurses using Caption Guidance software acquired TTEs of sufficient quality to make qualitative assessments of RV size and function in greater than 80% of cases for patients with implanted EP devices, and greater than 90% in patients without them. Sonographers performed similarly: descriptively, there was no significant difference between nurse- and sonographer-acquired scans for all comparisons (i.e., patients with EP devices, patients without EP devices, and overall).
The results of this analysis indicate that the software performs well in patients with implanted EP devices, supporting the overall conclusions of the pivotal study demonstrating that Caption Guidance can guide novice ultrasound users to obtain diagnostic-quality TTEs generally. These findings indicate that the deep learning technology used in the software is robust to large visual artifacts, such as pacemaker and ICD leads.
"Cardiac leads can produce dramatic visual artifacts in ultrasound images, which could confuse an algorithm that otherwise works well on typical images, making this research an excellent 'stress test' of the technology," said Sam Surette, first author on the abstract and Head of Regulatory Affairs and Quality Assurance at Caption Health. "Our findings are exciting because they provide further evidence that the core artificial intelligence in our software generalizes well to these patients."
Two additional analyses from the same study were presented earlier this month at the American Society of Echocardiography (ASE) 2020 Virtual Experience.
Caption Guidance works in tandem with Caption Interpretation™ as part of the Caption AI platform. Caption Interpretation applies end-to-end deep learning to automatically select the best clips from ultrasound exams, perform quality assurance and produce an accurate ejection fraction (EF) measurement. The technology incorporates three ultrasound views into its fully automated EF calculation: apical 4-chamber (AP4), apical 2-chamber (AP2) and—an industry first—parasternal long-axis (PLAX). An innovative update to Caption Interpretation was recently cleared by the FDA, marking the company's third clearance in the past six months.
About Caption Health
Caption Health was founded on a simple but powerful concept: What if we could use technology to emulate the expertise of highly trained medical experts and put that ability into the hands of every care provider? Our vision is to move specialty diagnostics and decision-making to the point of care for every patient exam. We are delivering AI systems that empower healthcare providers with new capabilities to acquire and interpret ultrasound exams. For more information, visit captionhealth.com.
Media Contact:
[email protected]
SOURCE Caption Health
![](https://rt.prnewswire.com/rt.gif?NewsItemId=SF09219&Transmission_Id=202008290700PR_NEWS_USPR_____SF09219&DateId=20200829)
Share this article