The Benefits Of AI In Cardiovascular Imaging
Artificial intelligence (AI) has increasingly impacted imaging modalities such as echocardiography, computed tomography (CT), and magnetic resonance imaging (MRI). The potential of this form of computer science has grown concurrently with the availability of large datasets, offering ways to improve patient care throughout the imaging chain.
In this article, we will put a spotlight on AI’s role in cardiac imaging, including the impact it’s had on important imaging modalities and how crucial it is for future development. How far has AI come? And which challenges lie ahead?
What is artificial intelligence?
AI is a type of computer science that simulates human intelligence processes. It has a wide range of applications; from machine learning to expert systems and speech recognition. AI uses both specialized hardware and software and a variety of programming languages, including Java and Python.
With the ability to analyze large datasets for patterns and correlations, AI can learn to identify, describe, and predict. The three cognitive skills involved in AI are:
- Learning – the acquisition of data and creation of an algorithm (rules system) that turns the data into information that can be actioned.
- Reasoning – programming that ensures the selection of the correct algorithm to achieve the desired outcome.
- Self-correction – the ongoing fine-tuning of algorithms in order to obtain the most accurate results.
Benefits of artificial intelligence
So why is AI being increasingly implemented in so many spheres? Let’s run through some of the main benefits of AI technology:
- Up efficiency: AI technology allows tasks to be completed faster, increasing efficiency. Whether it is getting through small tasks more quickly, or completing complex tasks in a shorter time than the mind is capable of; AI is pushing back the boundaries of what is possible and freeing up human time.
- 24/7 availability: humans need to sleep, so working for 24 hours a day is impossible. Not only does AI mimic and sometimes surpass the human mind, but it is also able to work around the clock, with no breaks required.
- Reduce error rates: because AI systems can work 24/7 without getting tired, they can also process more information without losing concentration, giving them the upper hand on humans again. AI cuts out human error.
- Faster decision-making: used with the right technology and programming, AI can enable machines to make faster decisions than humans, with no emotional influence. This delivers accurate results faster.
AI’s role in cardiovascular imaging
AI has excelled in cardiovascular imaging, and its potential is still being realized. AI has reduced human error in cardiovascular imaging applications and saved time in the clinical workflow. A major development has been AI’s ability to automatically segment cardiac structures, and there is the belief that in the future AI may be able to support and enable disease detection, prognosis, and decision making by expanding the value of diagnostic images.
For more information, please read our article on the potential of AI in cardiovascular medicine and imaging.
AI in echocardiography
Let’s hone in on AI’s role in echocardiography and its benefits for the modality. As the most accessible imaging modality in cardiology, echocardiography depends largely on the human interpretation of data by operators. AI has demonstrated its potential to replace human operators to some extent and reduce the intensive training required.
One study showed that a deep learning (DL) algorithm was able to perform view classification with the same accuracy as a board-certified echocardiographer. Other research used a DL algorithm to detect wall motion abnormalities with a similar AUC to that produced by the cardiologists and sonographer readers.
AI in cardiac CT
It is expected that AI will increasingly be integrated into the workflow of cardiac CT, including coronary computed tomography angiography (CCTA). AI can benefit cardiac CT by offering increased accuracy to coronary artery disease management with a quantitative-based approach. It is anticipated that wider implementation will depend on the validation of algorithms in good medical practice.
A DL method to obtain CT images with reduced radiation doses has been validated, and a DL approach has also been used to calculate calcium scores from regular coronary CT angiography (CTA), enabling a reduction of radiation exposure.
AI in cardiac MRI (CMR)
Machine learning (ML) algorithms have been implemented in all aspects of the cardiac MRI imaging workflow, including undersampled image acquisition, automated analysis, and the post-processing and development of predictive models.
AI has been used for cardiac structure and infarct tissue segmentation, including the segmentation of the right and left ventricular endo- and epicardium to calculate cardiac mass and function parameters from a number of datasets. Another study used a convolutional neural network to automate LV segmentation in all short-axis slices and phases in publicly available datasets. Texture analysis has also been helped by ML algorithms, selecting the most important cine-image-derived texture features to distinguish between patients with myocardial infarction and control subjects.
The drawback of AI in cardiovascular imaging
What are the potential drawbacks and challenges of applying AI in cardiovascular imaging? There is an ever-present need to demonstrate that AI can enable a higher quality of care that improves patient outcomes in a cost-effective manner. And with the significant volumes of data being used in AI applications, there is the responsibility to secure patient information – when being stored or extracted from electronic health records. Another door of opportunity which is yet to be fully unlocked is improved comparability of various imaging modes; this could lead to enhanced diagnosis and prognosis with AI
Conclusion
While the technology’s potential is still being explored, AI has already proven important for cardiac imaging, demonstrating that it can equal, if not exceed, a human operator in some roles. AI is being used in almost all imaging modalities and throughout healthcare. We've only touched the surface of AI’s impact in this article, and despite the ongoing challenges of cost-effectiveness and standardization, the expanding possibilities of AI’s benefits to clinical decision-making are exciting. To learn more about the future of AI, please read our in-depth article on Cardiac Imaging in 2040.
Sources:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7350824/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6395045/
https://pubmed.ncbi.nlm.nih.gov/31103590/
https://www.hindawi.com/journals/bmri/2020/6649410/
https://pubmed.ncbi.nlm.nih.gov/28574346/
https://pubmed.ncbi.nlm.nih.gov/29361481/
https://pubmed.ncbi.nlm.nih.gov/28437634/
https://www.frontiersin.org/articles/10.3389/fcvm.2021.818765/full