Caption Health’s artificial intelligence has been described by Time Magazine as one of the best inventions of 2021. The software works with a new generation of wearable ultrasound machines to take a view of the heart and help a doctor examine it for possible problems.
For decades, echocardiograms have helped medical professionals diagnose cardiovascular diseases such as rheumatic heart disease by examining the structure of the heart and surrounding blood vessels, analyzing how blood flows through them, and assessing the heart’s pumping chambers.
An echocardiogram (or echo) is a type of ultrasound scan used to examine the heart and nearby blood vessels. It uses a small probe to emit high-frequency sound waves that create echoes as they bounce off different parts of the body.
These echoes are picked up by the probe and converted into a moving image on a monitor during the scan.
A heart specialist or cardiologist, or any doctor who thinks a patient has a heart problem, may request an echocardiogram.
The test is usually performed in a hospital or clinic by a cardiologist, cardiac physiologist, or a trained technician called a sonographer.
Unfortunately, many developing countries lack ultrasound equipment and trained technicians. This fallout can potentially be bridged through the intervention of machine learning software.
According to a study published in the International Journal of Research in Medical Sciences, there is an increasing trend in the burden of cardiovascular disease in Nigeria, driven by increasing rural-urban migration and socio-economic changes. World Health Organization data released in 2020 showed that the number of deaths from coronary artery disease in Nigeria reached 61,374. Heart disease contributes to millions of deaths worldwide, but early symptoms such as fatigue or shortness of breath are easily mistaken for normal signs of aging, particularly in those over 65 years of age.
While traditional ultrasounds used in echocardiography are highly effective diagnostic tools, particularly in detecting cardiac (as well as pulmonary) disease, they have been found to have certain underlying challenges. Scientists observe that performing the traditional ultrasound scan is associated with unnatural hand-eye coordination and unintuitive images.
Additionally, technologists say the software the ultrasound uses does not provide instructions for moving the transducer to capture an image, or real-time feedback on the quality of the images captured.
In addition, traditional ultrasound has been found to be limited by poor image quality, misdiagnosis, repeated studies, and inconsistent interpretation. This is why the Caption Health AI echocardiogram was invented to see the true value and benefits of ultrasound.
With Caption AI technology, medical professionals can easily capture high-quality ultrasound images using AI, which will help them scan and capture the process in real time and calculate ejection fraction – a key indicator of heart health.
This technology can help detect signs of diseases like heart failure in at-risk patients in physician’s offices, at home and in alternative care settings, potentially preventing hospitalizations and supporting improved clinical outcomes.
In a YouTube video, a healthcare professional identified simply as Sarah demonstrated how to perform an echocardiogram with Caption AI. To begin, Sarah chose a protocol to follow.
Caption AI allows users to create custom protocols with up to 10 default cardiac views based on workflow needs or patient indications.
She launched the parasternal long axis view and used the diagram on the bottom right of the device to position the transducer.
Above that was a reference image that showed her what the current view she was looking for should look like.
Following the prompt at the top of the monitor, she performed specific transducer movements to optimize the image.
On the left, the quality meter provided some information on how close she was to capturing a diagnostic-quality image.
As Sarah neared optimal vision, the gauge rose. Once the software recognized a diagnostic-quality image, the meter turned green and the clip was automatically recorded without the need to press a button.
Next she switched to the apical four and the apical two chamber views and followed the real-time guidance and made more fine movements to improve the image.
Adding multiple views automatically displays the ejection fraction measurement.
Caption AI keeps track of the best frames seen during scanning, allowing users to save the best clip and move on.
The captioned AI report page displayed the final ejection fraction measurement and all three views were mapped with the appropriate clinical guidelines to show if the heart being examined had an abnormality or health condition.
A UK-based Nigerian cardiologist, Dr. Livinus Ede, told Sunday PUNCH that some of the admirable benefits of AI in healthcare are the elimination of association errors and the reduction of patient exposure to radiation.
“AI is the future of imaging. Not only does it help in creating improved cross-sectional images, but it also eliminates many associated operator errors.
“Another important aspect is the ability to generate real-time feedback, leading to flexibility in acquisition, thus reducing time, cost and recurring radiation exposure,” he said.
From the United States of America, where Caption AI was first developed, its use has spread to Europe.
Steve Cashman, Chief Executive Officer of Caption Health reportedly said, “Our strategic global expansion is about more than market access, it’s about patient access.”
Another UK-based Nigerian cardiologist, Dr. Isaiah Osin, told our correspondent that Nigeria suffers from a lack of trained hands in the field of echocardiography and would benefit greatly from this AI contribution to healthcare.
He said: “AI would definitely benefit Nigeria’s healthcare system in many ways and for many reasons. In the field of echocardiography, Nigeria has a shortage of trained hands and considering what AI can do means greater access to quality imaging and diagnostics with fewer training requirements for healthcare providers.
“Even for those trained in echocardiography, this means a more accurate diagnosis with less human error. Therefore, AI-enabled ultrasound is a step in the right direction for Nigeria.”
Nigeria’s rapidly growing population is overwhelming the understaffed and underfunded healthcare system, but machine learning and predictive analytics, particularly in the area of disease diagnosis, will improve access to treatment and the integration of prevention and care into Nigeria’s healthcare system.
According to a study published in the European Journal of Radiology, the diagnostic power of conventional ultrasound is inevitably reduced due to the inherent characteristic of high operator dependency on the United States.
In contrast, AI excels at automatically recognizing complex patterns and providing a quantitative assessment of image data, showing high potential to help physicians obtain more accurate and reproducible results.
A medical sonographer in Benin City, Mr. Yinka Awopetu, pointed out that the invention of AI would not stop at improving heart disease diagnosis, but would extend to other health care operations.
He said: “AI is a branch of computer science that includes machine learning. AI can thus be recognized as any device designed to mimic human cognitive processes, capable of cognitively solving complex human questions or problems.
“On the other hand, ultrasound in its entirety has to do with proper hands-on training in all of its specialties, which include cardiovascular, pelvic, abdominal, ocular, thoracic and soft tissue.
“Therefore, ultrasound image analysis will enter a tremendous new era in the next few years if this dependency between operator, scanner and patient can be significantly improved.
“In the next few years, AI will be a great adopted practice as the ultrasound practice is self-dependent and AI is a digital health system, so the Caption Health AI will revolutionize ultrasound imaging.”
dr For his part, Olaleye Oluwasanmi, a medical practitioner at the Aviation Medical Clinics in Lagos, emphasized the need for AI to change the phase of healthcare in Nigeria.
However, he said it would be better if the invention were used alongside traditional diagnostic tools to consider the human contribution in disease diagnosis.
“The invention will help the state of healthcare in Nigeria when used in conjunction with the existing diagnostic tools such as ultrasound and human intervention in diagnosis.
“No matter how fantastic a health AI is, human input is always needed. The emotional factor and interactive effect that the human operator has cannot be completely replaced by an AI.
“While this is a welcome development, I will suggest that it be used alongside what we currently have, rather than replacing it entirely,” he said.