Automation, machine learning, and artificial intelligence in echocardiography: A brave new world

From Digital Culture & Society

Jump to: navigation, search

Chow, Chi-Ming., Gandhi, Sumeet., Mosleh Wassim., Shen, Joshua. (2018). “Automation, machine learning, and artificial intelligence in echocardiography: A brave new world.” Wiley Echocardiography, 35(9), 1402-1417.

Find article online:

DOI: 10.1111/echo.14086

Contents

[edit] Context

This article discusses how automation, machine learning, and artificial intelligence have been utilized as tools in echocardiography to assist medical physicians in their work. As technological practices have become advanced over time, the scope of those practices are now able to reach different sectors of work. As discussed in this article, the realm of medical work has the benefit of adopting deep learning to assist medical physicians in their work. Although the usage of machine learning in the medical field is still in its infancy, its potential allows for the training of data sets to go beyond what traditional statistical reasoning can handle. These methods of deep learning when applied to large image archives will be able to recognize complex patterns that helps to establish a better understanding of the prognosis of cardiac diseases. The purpose of this journal article was to describe the methods of deep learning to discuss how it will impact echocardiography in the future.

[edit] Overview

This article begins by discussing the boundaries of deep learning by explaining how they can aid clinical decision-making. The authors explain the concepts of artificial intelligence and define terms associated with deep learning (Chow p. 1403). Practices of deep learning are seen as suitable for clinical use because of how machine learning acts as an artificial neural network. An artificial neural network assists in making targeted decisions based upon the weights of neurons found in a human brain, in order to discover these neurons the process of machine learning is used. To apply this data, convolution neural network is used to identify echocardiographic views which can assist in the prognosis of cardiac diseases through echocardiography. In order to support their research the authors have taken an abundance of other studies that pertain to automation in echocardiography to summarize findings between the fields of research (Chow p. 1405). The main portion of the article is an explanation of how automation can be applied to echocardiographic data such as knowledge-based identification of cardiac chamber shapes. A limitation in their image collection was discovered when the software used did not produce quality images (Chow 1413). Echocardiographic analysis acts as a component of decision-making and management in patients with valve disease. The authors conclude their study by deeming that current machines are not able to replace human experts in echocardiography interpretation and that the practice with machines will need further development.

[edit] Strengths and Weaknesses

The main strength of this article is how it thoroughly articulates how artificial intelligence, machine learning, and automation have proficient usage in the medical field. Each practice of deep learning has its own section within the article where it is explained as to how it is relevant to echocardiographic imagery. The second strength of this article is how the field of automation in cardiography is showcased. The authors have incorporated an extensive amount of other research data to correlate their discussion in the practices of deep learning within echocardiographic imaging. The research data is summarized within tables in the article that explain the main finding from the author of each research study.

The weakness of this article is how it does not compute the statistics or methodology of how effective machine learning and automation is within relation to echocardiographic imagery. This is a weakness due to how it is discussed how deep learning can benefit this field but the results of how it benefits aren’t necessarily outlined within the article. While the authors go into extensive detail as to how various machine-learning algorithms and automation is incorporated into echocardiographic imagery, the measurement of how beneficial computer science in the medical field isn’t necessarily detailed.

[edit] Assessment

In conclusion, this article provides an excellent explanation of how machines can be utilized in the medical field. The authors utilize the work of their peers to assist their research while also explaining how each sector of deep learning can help contribute towards medical physicians in their line of work. Nonetheless, the authors are not biased in their work as they also delineate how their are limitations of automation and machine learning in echocardiographic imagery. Even though machine errors are probable, AI has the ability to optimize productivity and push the boundaries of technological discovery in the field of medicine.


--Bl14hl 21:06, 19 March 2019 (EDT)

Personal tools
Bookmark and Share