<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.1 20151215//EN" "https://jats.nlm.nih.gov/publishing/1.1/JATS-journalpublishing1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" dtd-version="1.1" specific-use="sps-1.9" article-type="editorial" xml:lang="en">
<front>
<journal-meta>
<journal-id journal-id-type="publisher-id">abcimg</journal-id>
<journal-title-group>
<journal-title>ABC Imagem Cardiovascular</journal-title>
<abbrev-journal-title abbrev-type="publisher">Arq Bras Cardiol: Imagem cardiovasc.</abbrev-journal-title></journal-title-group>
<issn pub-type="epub">2675-312X</issn>
<publisher>
<publisher-name>Sociedade Brasileira de Cardiologia</publisher-name></publisher>
</journal-meta>
<article-meta>
<article-id pub-id-type="doi">10.36660/abcimg.20240095i</article-id>
<article-id pub-id-type="publisher-id">abcimg.20240095i</article-id>
<article-categories>
<subj-group subj-group-type="heading">
<subject>Editorial</subject>
</subj-group>
</article-categories>
<title-group>
<article-title>Artificial Intelligence in Echocardiography: The Future of Precision Diagnosis</article-title>
</title-group>
<contrib-group>
<contrib contrib-type="editor">
<contrib-id contrib-id-type="orcid">0000-0002-2634-4632</contrib-id>
<name><surname>Otto</surname><given-names>Maria Estefânia Bosco</given-names></name>
<xref ref-type="aff" rid="aff1"><sup>1</sup></xref>
<xref ref-type="aff" rid="aff2"><sup>2</sup></xref>
<xref ref-type="corresp" rid="c1"/></contrib>
<aff id="aff1">
<label>1</label>
<institution content-type="orgname">Instituto de Cardiologia do Distrito Federal</institution>
<addr-line>
<named-content content-type="city">Brasília</named-content>
<named-content content-type="state">DF</named-content>
</addr-line>
<country country="BR">Brazil</country>
<institution content-type="original">Instituto de Cardiologia do Distrito Federal (ICDF) Brasília, DF – Brazil</institution>
</aff>
<aff id="aff2">
<label>2</label>
<institution content-type="orgname">DF Star</institution>
<addr-line>
<named-content content-type="city">Brasília</named-content>
<named-content content-type="state">DF</named-content>
</addr-line>
<country country="BR">Brazil</country>
<institution content-type="original">DF Star, Brasília, DF – Brazil</institution>
</aff>
</contrib-group>
<author-notes>
<corresp id="c1"><label>Mailing Address:</label> Maria Estefania Otto • Instituto de Cardiologia do Distrito Federal (ICDF). SQSW 301, Bloco F, Apto 508. Postal code: <postal-code>70673-106</postal-code>. Brasília, DF – Brazil. E-mail: <email>mariaestefaniaotto@gmail.com</email></corresp>
</author-notes>
<pub-date publication-format="electronic" date-type="pub">
<day>22</day>
<month>11</month>
<year>2024</year></pub-date>
<pub-date publication-format="electronic" date-type="collection">
<year>2024</year></pub-date>
<volume>37</volume>
<issue>4</issue>
<elocation-id>e20240095</elocation-id>
<permissions>
<license license-type="open-access" xlink:href="https://creativecommons.org/licenses/by/4.0/" xml:lang="en">
<license-p>This is an open-access article distributed under the terms of the Creative Commons Attribution License</license-p>
</license>
</permissions>
<kwd-group xml:lang="en">
<title>Keywords</title>
<kwd>Artificial Intelligence</kwd>
<kwd>Echocardiography</kwd>
<kwd>Diagnosis</kwd>
</kwd-group>
<counts>
<fig-count count="0"/>
<table-count count="2"/>
<equation-count count="0"/>
<ref-count count="16"/>
</counts>
</article-meta>
</front>
<body>
<p>The integration of artificial intelligence (AI) into echocardiography ushers in a transformative era in cardiovascular imaging, offering greater diagnostic accuracy, workflow efficiency, and data integration. The emergence of AI-based technologies in various areas of cardiovascular imaging is redefining not only our field of expertise, but also the interaction between diagnostic medicine and clinical practice. This editorial aims to provide basic definitions of terms used in AI, summarized in <xref ref-type="table" rid="t1">Table 1</xref>,<sup><xref ref-type="bibr" rid="B1">1</xref></sup> and explore the growing role of AI in echocardiography, with emphasis on its applications in systolic and diastolic functions, mitral and aortic valve diseases, and the increasing use of point-of-care ultrasound (POCUS). Furthermore, it provides a final reflection on the advantages and disadvantages of using AI in diagnostic medicine.</p>
<table-wrap id="t1">
<label>Table 1</label>
<caption><title>Definition of terms associated with AI.</title></caption>
<table frame="hsides" rules="groups">
<colgroup width="33%">
<col/>
<col/>
<col/>
</colgroup>
<thead style="border-top: thin solid; border-bottom: thin solid; border-color: #000000">
<tr style="background-color:#C58874">
<th align="left" valign="middle">Term</th>
<th align="center" valign="middle">Abbreviation</th>
<th align="center" valign="middle">Concept</th>
</tr>
</thead>
<tbody style="border-bottom: thin solid; border-color: #000000">
<tr>
<td align="left" valign="middle">Artificial Intelligence</td>
<td align="center" valign="middle">AI</td>
<td align="center" valign="middle">Technology that enables computer systems to imitate human cognitive functions, such as learning, reasoning, and problem-solving.</td>
</tr>
<tr style="background-color:#E8CCBF">
<td align="left" valign="middle">Convolutional Neural Network</td>
<td align="center" valign="middle">CNN</td>
<td align="center" valign="middle">A specialized architecture for analyzing visual images, within the category of deep neural networks, using convolutional layers to capture spatial patterns.</td>
</tr>
<tr>
<td align="left" valign="middle">Deep Learning</td>
<td align="center" valign="middle">DL</td>
<td align="center" valign="middle">A subfield of ML that uses multilayer neural networks to learn from large data sets, especially effective for complex tasks, such as image recognition.</td>
</tr>
<tr style="background-color:#E8CCBF">
<td align="left" valign="middle">Deep Neural Network</td>
<td align="center" valign="middle">DNN</td>
<td align="center" valign="middle">A DL framework designed to analyze complex data patterns, using multiple layers of processing to improve prediction accuracy.</td>
</tr>
<tr>
<td align="left" valign="middle">Large Language Model</td>
<td align="center" valign="middle">LLM</td>
<td align="center" valign="middle">Advanced DL models, such as GPT (exactly the model in GPT chat), trained on large volumes of textual data, effective in processing and producing natural language.</td>
</tr>
<tr style="background-color:#E8CCBF">
<td align="left" valign="middle">Machine Learning</td>
<td align="center" valign="middle">ML</td>
<td align="center" valign="middle">A branch of AI that develops algorithms to train computers to analyze data and make predictions, enabling autonomous learning from large volumes of data, such as recognizing an echocardiogram location of acoustic window.</td>
</tr>
<tr>
<td align="left" valign="middle">Natural Language Processing</td>
<td align="center" valign="middle">NLP</td>
<td align="center" valign="middle">A field of AI that enables machines to understand and interpret human language, crucial for such applications as language translation and sentiment analysis.</td>
</tr>
</tbody>
</table>
<table-wrap-foot>
<fn id="TFN1"><p>Adapted from Lüscher et al.<sup><xref ref-type="bibr" rid="B1">1</xref></sup></p></fn>
</table-wrap-foot>
</table-wrap>
<sec>
<title>AI and the Systolic Function</title>
<p>Ejection fraction (EF) is an essential parameter used to assess the systolic function; however, it has been historically subjected to interobserver variability and the level of experience of the examiner.<sup><xref ref-type="bibr" rid="B2">2</xref></sup> AI-based algorithms present solutions for EF, providing automated and reproducible calculations.<sup>3–5</sup> A pioneering study published by He et al.<sup><xref ref-type="bibr" rid="B6">6</xref></sup> evaluated a deep learning (DL) system comprised of convolutional neural networks (CNNs), called EchoNet-Dynamic flow, and compared EF assessments performed by AI with measurements performed by experienced sonographers. The reference cardiologists who checked these results were unable to differentiate between the sonographer and AI. Furthermore, the results demonstrated that the assessment by AI was not inferior to those performed by experts, with a difference of less than 5% in variability. Other advantages of this system were the speed of measurement and the possibility of averaging up to five heartbeats, thus improving the accuracy of the result in atrial fibrillation.</p>
<p>Another parameter widely used to assess systolic function in echocardiography is the Global Longitudinal Strain (GLS) analysis,<sup><xref ref-type="bibr" rid="B7">7</xref></sup> which not only aids in the diagnosis of cardiac function, but also reveals characteristic patterns of certain diseases, such as amyloidosis.<sup><xref ref-type="bibr" rid="B7">7</xref></sup> The GLS technique itself requires machine learning (ML) training of the equipment to identify acoustic window incidences and outline edges. However, AP techniques by CNR were able to assess GLS more quickly and with less variability when compared to a commercially established workstation used in echocardiography laboratories.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> Recently, Kwan et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> evaluated a CNR system used to assess GLS with images produced by equipment from two different companies, concluding that, in addition to having less variability and good reproducibility, this system was able to identify GLS patterns and differentiate between athlete&apos;s hypertrophy, amyloidosis, and hypertrophic cardiomyopathy. However, it important to understand that the quality of the images obtained for both GLS and EF assessments must be good, a factor that cannot yet be achieved by AI.</p>
</sec>
<sec>
<title>AI and the Diastolic Function</title>
<p>Assessment of left ventricular diastolic function plays an essential role in the diagnosis and prognosis of several heart diseases, especially heart failure with preserved EF. However, the multiple algorithms and complexity of echocardiographic parameters used to assess diastolic function make it difficult to adequately classify approximately 20% of all patients, even when experienced professionals perform the analysis.<sup><xref ref-type="bibr" rid="B10">10</xref></sup> In this group, classified as having indeterminate diastolic function, it is not possible to define the presence or absence of elevated filling pressures, an essential parameter in determining prognosis and aiding in treatment.<sup><xref ref-type="bibr" rid="B11">11</xref></sup></p>
<p>In this scenario, electrocardiogram (ECG) data, detectable primarily through AI AP, emerges as a promising solution to detect patients with increased filling pressures. Lee et al.<sup><xref ref-type="bibr" rid="B12">12</xref></sup> analyzed more than 90,000 ECGs performed by AI in patients who underwent evaluation for diastolic function with a maximum difference of 14 days. By incorporating ECG into echocardiogram data, the accuracy for detecting elevated filling pressure was 91%, rivaling that of traditional echocardiographic assessments. Thus, the use of AI-assessed ECG can complement echocardiogram data, especially in situations where the number of parameters is quite large and where there is a lack of dichotomization. The ability to predict diastolic dysfunction and elevated filling pressures from a simple ECG underscores the role of AI in expanding diagnostic capacity in resource-limited settings and has now been incorporated into the echocardiogram report of the authors of the article.</p>
</sec>
<sec>
<title>AI in Mitral Valve Disease</title>
<p>Mitral regurgitation (MR) is highly prevalent in the population and is associated with a significant reduction in quality of life. Transthoracic echocardiography (TTE) plays an essential role in detecting the presence and quantification of MR. However, it is subject to significant interobserver variability regarding measurements of volume and regurgitant orifice. MR assessment by AI is challenging, since only a few images obtained during the examination contain the data that define severity. More recently, Long et al.<sup><xref ref-type="bibr" rid="B13">13</xref></sup> described an AP model capable of interpreting TTE video data and automatically classifying MR severity. This model integrates color Doppler video imaging, offering precise grading of MR on a scale ranging from mild to severe, with an accuracy of approximately 80%.</p>
<p>Furthermore, the predictive power of AI is highlighted by the development of risk stratification tools, such as the EuroSMR score,<sup><xref ref-type="bibr" rid="B14">14</xref></sup> designed to predict outcomes in patients undergoing transcatheter edge-to-edge repair for secondary CABG. This AI-derived tool has demonstrated robust performance in predicting mortality, aiding clinicians in making decisions together with the Heart Team. This score can be accessed online at www.eurosmr.com.</p>
</sec>
<sec>
<title>AI and Aortic Valve Disease</title>
<p>Aortic stenosis (AS) is a very prevalent and lethal valve disease, especially in elderly patients. Once again, the method of choice to assess AS is TTE, integrated with Doppler parameters and analysis of the left ventricular outflow tract. However, there is a need for numerous variables, which in addition to being time-consuming to analyze, are subject to intra- and interobserver variability. Fully automated AI systems have been developed to address these challenges. In the study by Krishna et al.,<sup><xref ref-type="bibr" rid="B5">5</xref></sup> AP algorithms with CNR were applied to a cohort of patients with different degrees of AS. The result was an excellent correlation, with measurements made by experienced cardiologists, with the advantage of substantially reducing analysis time.</p>
<p>The role of AI in the assessment of AS is not only limited to improving measurement accuracy, but it also extends to the identification of early markers of the disease. These technologies are especially promising in screening programs, where the rapid and accurate assessment of valve function is essential.</p>
</sec>
<sec>
<title>AI in POCUS</title>
<p>The introduction of AI in POCUS has opened up new possibilities for real-time diagnostic assessments, especially in low-resource settings. AI-enabled POCUS devices assist non-specialists by automating image acquisition and interpretation, reducing the learning curve. For example, AI can guide the operator in obtaining the correct acoustic windows, significantly reducing the learning curve in POCUS.<sup><xref ref-type="bibr" rid="B15">15</xref></sup></p>
<p>AI applications in POCUS are particularly beneficial in emergency scenarios, where rapid decision-making is essential. The use of AI can facilitate lung ultrasound for the consistent detection of congestion and pleural effusion. Furthermore, in rural and remote areas, where access to expert interpretation is limited, AI can provide real-time diagnostic support, making access to advanced technologies more democratic.<sup><xref ref-type="bibr" rid="B16">16</xref></sup></p>
</sec>
<sec>
<title>Reflections on AI in cardiovascular imaging</title>
<p>AI streamlines and simplifies image evaluation, optimizing the workflow in echocardiography laboratories. However, image acquisition is still performed manually, and for CNR analysis to be accurate, the images must be of high quality, making the human component indispensable. In addition, another limitation of AI is the need for a significant volume of images to properly train the CNR model.</p>
<p>We can reflect on two opposing points of view. According to Yuval Harari, AI is not a passive technology, but rather a system that learns autonomously, which can lead to a decrease in human empowerment, as described in his book <italic>Nexus: A Brief History of Information Networks</italic> (2024). By contrast, Kai Fu Lee, one of the leading scholars of AI since its inception and author of several works on the subject, argues that AI will not replace humans, but rather humans who use AI will replace those who do not use it.</p>
<sec sec-type="conclusions">
<title>Conclusion</title>
<p>AI is poised to play a central role in the future of echocardiography, improving diagnostic accuracy, increasing efficiency, and optimizing workflow in echocardiography laboratories. However, with these advancements also come challenges, including the need for robust validation in diverse populations and rigorous regulatory oversight. As the field advances, collaboration among AI developers, cardiovascular imaging experts, and regulators will be crucial to ensuring the safe and effective integration of AI into clinical practice.</p>
</sec>
</sec>
</body>
<back>
<ref-list>
<title>References</title>
<ref id="B1">
<label>1</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Lüscher</surname><given-names>TF</given-names></name>
<name><surname>Wenzl</surname><given-names>FA</given-names></name>
<name><surname>D’Ascenzo</surname><given-names>F</given-names></name>
<name><surname>Friedman</surname><given-names>PA</given-names></name>
<name><surname>Antoniades</surname><given-names>C</given-names></name>
</person-group>
<article-title>Artificial Intelligence in Cardiovascular Medicine: Clinical Applications</article-title>
<source>Eur Heart J</source>
<year>2024</year>
<volume>45</volume>
<issue>40</issue>
<fpage>4291</fpage>
<lpage>4304</lpage>
<pub-id pub-id-type="doi">10.1093/eurheartj/ehae465</pub-id>
</element-citation>
<mixed-citation>Lüscher TF, Wenzl FA, D’Ascenzo F, Friedman PA, Antoniades C. Artificial Intelligence in Cardiovascular Medicine: Clinical Applications. Eur Heart J. 2024;45(40):4291-304. doi: 10.1093/eurheartj/ehae465.</mixed-citation>
</ref>
<ref id="B2">
<label>2</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Lang</surname><given-names>RM</given-names></name>
<name><surname>Badano</surname><given-names>LP</given-names></name>
<name><surname>Mor-Avi</surname><given-names>V</given-names></name>
<name><surname>Afilalo</surname><given-names>J</given-names></name>
<name><surname>Armstrong</surname><given-names>A</given-names></name>
<name><surname>Ernande</surname><given-names>L</given-names></name>
<etal/>
</person-group>
<article-title>Recommendations for Cardiac Chamber Quantification by Echocardiography in Adults: An Update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging</article-title>
<source>J Am Soc Echocardiogr</source>
<year>2015</year>
<volume>28</volume>
<issue>1</issue>
<fpage>1</fpage>
<lpage>39.e14</lpage>
<pub-id pub-id-type="doi">10.1016/j.echo.2014.10.003</pub-id>
</element-citation>
<mixed-citation>Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, et al. Recommendations for Cardiac Chamber Quantification by Echocardiography in Adults: An Update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr. 2015;28(1):1-39.e14. doi: 10.1016/j.echo.2014.10.003.</mixed-citation>
</ref>
<ref id="B3">
<label>3</label>
<element-citation publication-type="book">
<person-group person-group-type="author">
<name><surname>Ouyang</surname><given-names>D</given-names></name>
</person-group>
<chapter-title>EchoNet-RCT: Blinded, Randomized Controlled Trial of Sonographer vs</chapter-title>
<source>Artificial Intelligence Assessment of Cardiac Function</source>
<publisher-loc>Barcelona</publisher-loc>
<publisher-name>ESC Congress</publisher-name>
<year>2022</year>
</element-citation>
<mixed-citation>Ouyang D. EchoNet-RCT: Blinded, Randomized Controlled Trial of Sonographer vs. Artificial Intelligence Assessment of Cardiac Function. Barcelona: ESC Congress; 2022.</mixed-citation>
</ref>
<ref id="B4">
<label>4</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Ouyang</surname><given-names>D</given-names></name>
<name><surname>He</surname><given-names>B</given-names></name>
<name><surname>Ghorbani</surname><given-names>A</given-names></name>
<name><surname>Yuan</surname><given-names>N</given-names></name>
<name><surname>Ebinger</surname><given-names>J</given-names></name>
<name><surname>Langlotz</surname><given-names>CP</given-names></name>
<etal/>
</person-group>
<article-title>Video-based AI for Beat-to-beat Assessment of Cardiac Function</article-title>
<source>Nature</source>
<year>2020</year>
<volume>580</volume>
<issue>7802</issue>
<fpage>252</fpage>
<lpage>256</lpage>
<pub-id pub-id-type="doi">10.1038/s41586-020-2145-8</pub-id>
</element-citation>
<mixed-citation>Ouyang D, He B, Ghorbani A, Yuan N, Ebinger J, Langlotz CP, et al. Video-based AI for Beat-to-beat Assessment of Cardiac Function. Nature. 2020;580(7802):252-6. doi: 10.1038/s41586-020-2145-8.</mixed-citation>
</ref>
<ref id="B5">
<label>5</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Krishna</surname><given-names>H</given-names></name>
<name><surname>Desai</surname><given-names>K</given-names></name>
<name><surname>Slostad</surname><given-names>B</given-names></name>
<name><surname>Bhayani</surname><given-names>S</given-names></name>
<name><surname>Arnold</surname><given-names>JH</given-names></name>
<name><surname>Ouwerkerk</surname><given-names>W</given-names></name>
<etal/>
</person-group>
<article-title>Fully Automated Artificial Intelligence Assessment of Aortic Stenosis by Echocardiography</article-title>
<source>J Am Soc Echocardiogr</source>
<year>2023</year>
<volume>36</volume>
<issue>7</issue>
<fpage>769</fpage>
<lpage>777</lpage>
<pub-id pub-id-type="doi">10.1016/j.echo.2023.03.008</pub-id>
</element-citation>
<mixed-citation>Krishna H, Desai K, Slostad B, Bhayani S, Arnold JH, Ouwerkerk W, et al. Fully Automated Artificial Intelligence Assessment of Aortic Stenosis by Echocardiography. J Am Soc Echocardiogr. 2023;36(7):769-77. doi: 10.1016/j.echo.2023.03.008.</mixed-citation>
</ref>
<ref id="B6">
<label>6</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>He</surname><given-names>B</given-names></name>
<name><surname>Kwan</surname><given-names>AC</given-names></name>
<name><surname>Cho</surname><given-names>JH</given-names></name>
<name><surname>Yuan</surname><given-names>N</given-names></name>
<name><surname>Pollick</surname><given-names>C</given-names></name>
<name><surname>Shiota</surname><given-names>T</given-names></name>
<etal/>
</person-group>
<article-title>Blinded, Randomized Trial of Sonographer versus AI Cardiac Function Assessment</article-title>
<source>Nature</source>
<year>2023</year>
<volume>616</volume>
<issue>7957</issue>
<fpage>520</fpage>
<lpage>524</lpage>
<pub-id pub-id-type="doi">10.1038/s41586-023-05947-3</pub-id>
</element-citation>
<mixed-citation>He B, Kwan AC, Cho JH, Yuan N, Pollick C, Shiota T, et al. Blinded, Randomized Trial of Sonographer versus AI Cardiac Function Assessment. Nature. 2023;616(7957):520-4. doi: 10.1038/s41586-023-05947-3.</mixed-citation>
</ref>
<ref id="B7">
<label>7</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Almeida</surname><given-names>ALC</given-names></name>
<name><surname>Melo</surname><given-names>MDT</given-names></name>
<name><surname>Bihan</surname><given-names>DCSL</given-names></name>
<name><surname>Vieira</surname><given-names>MLC</given-names></name>
<name><surname>Pena</surname><given-names>JLB</given-names></name>
<name><surname>Del Castillo</surname><given-names>JM</given-names></name>
<etal/>
</person-group>
<article-title>Position Statement on the Use of Myocardial Strain in Cardiology Routines by the Brazilian Society of Cardiology&apos;s Department Of Cardiovascular Imaging - 2023</article-title>
<source>Arq Bras Cardiol</source>
<year>2023</year>
<volume>120</volume>
<issue>12</issue>
<elocation-id>e20230646</elocation-id>
<pub-id pub-id-type="doi">10.36660/abc.20230646</pub-id>
</element-citation>
<mixed-citation>Almeida ALC, Melo MDT, Bihan DCSL, Vieira MLC, Pena JLB, Del Castillo JM, et al. Position Statement on the Use of Myocardial Strain in Cardiology Routines by the Brazilian Society of Cardiology&apos;s Department Of Cardiovascular Imaging - 2023. Arq Bras Cardiol. 2023;120(12):e20230646. doi: 10.36660/abc.20230646.</mixed-citation>
</ref>
<ref id="B8">
<label>8</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Salte</surname><given-names>IM</given-names></name>
<name><surname>Østvik</surname><given-names>A</given-names></name>
<name><surname>Smistad</surname><given-names>E</given-names></name>
<name><surname>Melichova</surname><given-names>D</given-names></name>
<name><surname>Nguyen</surname><given-names>TM</given-names></name>
<name><surname>Karlsen</surname><given-names>S</given-names></name>
<etal/>
</person-group>
<article-title>Artificial Intelligence for Automatic Measurement of Left Ventricular Strain in Echocardiography</article-title>
<source>JACC Cardiovasc Imaging</source>
<year>2021</year>
<volume>14</volume>
<issue>10</issue>
<fpage>1918</fpage>
<lpage>1928</lpage>
<pub-id pub-id-type="doi">10.1016/j.jcmg.2021.04.018</pub-id>
</element-citation>
<mixed-citation>Salte IM, Østvik A, Smistad E, Melichova D, Nguyen TM, Karlsen S, et al. Artificial Intelligence for Automatic Measurement of Left Ventricular Strain in Echocardiography. JACC Cardiovasc Imaging. 2021;14(10):1918-28. doi: 10.1016/j.jcmg.2021.04.018.</mixed-citation>
</ref>
<ref id="B9">
<label>9</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Kwan</surname><given-names>AC</given-names></name>
<name><surname>Chang</surname><given-names>EW</given-names></name>
<name><surname>Jain</surname><given-names>I</given-names></name>
<name><surname>Theurer</surname><given-names>J</given-names></name>
<name><surname>Tang</surname><given-names>X</given-names></name>
<name><surname>Francisco</surname><given-names>N</given-names></name>
<etal/>
</person-group>
<article-title>Deep Learning-derived Myocardial Strain</article-title>
<source>JACC Cardiovasc Imaging</source>
<year>2024</year>
<volume>17</volume>
<issue>7</issue>
<fpage>715</fpage>
<lpage>725</lpage>
<pub-id pub-id-type="doi">10.1016/j.jcmg.2024.01.011</pub-id>
</element-citation>
<mixed-citation>Kwan AC, Chang EW, Jain I, Theurer J, Tang X, Francisco N, et al. Deep Learning-derived Myocardial Strain. JACC Cardiovasc Imaging. 2024;17(7):715-25. doi: 10.1016/j.jcmg.2024.01.011.</mixed-citation>
</ref>
<ref id="B10">
<label>10</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Almeida</surname><given-names>JG</given-names></name>
<name><surname>Fontes-Carvalho</surname><given-names>R</given-names></name>
<name><surname>Sampaio</surname><given-names>F</given-names></name>
<name><surname>Ribeiro</surname><given-names>J</given-names></name>
<name><surname>Bettencourt</surname><given-names>P</given-names></name>
<name><surname>Flachskampf</surname><given-names>FA</given-names></name>
<etal/>
</person-group>
<article-title>Impact of the 2016 ASE/EACVI Recommendations on the Prevalence of Diastolic Dysfunction in the General Population</article-title>
<source>Eur Heart J Cardiovasc Imaging</source>
<year>2018</year>
<volume>19</volume>
<issue>4</issue>
<fpage>380</fpage>
<lpage>386</lpage>
<pub-id pub-id-type="doi">10.1093/ehjci/jex252</pub-id>
</element-citation>
<mixed-citation>Almeida JG, Fontes-Carvalho R, Sampaio F, Ribeiro J, Bettencourt P, Flachskampf FA, et al. Impact of the 2016 ASE/EACVI Recommendations on the Prevalence of Diastolic Dysfunction in the General Population. Eur Heart J Cardiovasc Imaging. 2018;19(4):380-6. doi: 10.1093/ehjci/jex252.</mixed-citation>
</ref>
<ref id="B11">
<label>11</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Robinson</surname><given-names>S</given-names></name>
<name><surname>Ring</surname><given-names>L</given-names></name>
<name><surname>Oxborough</surname><given-names>D</given-names></name>
<name><surname>Harkness</surname><given-names>A</given-names></name>
<name><surname>Bennett</surname><given-names>S</given-names></name>
<name><surname>Rana</surname><given-names>B</given-names></name>
<etal/>
</person-group>
<article-title>The Assessment of Left Ventricular Diastolic Function: Guidance and Recommendations from the British Society of Echocardiography</article-title>
<source>Echo Res Pract</source>
<year>2024</year>
<volume>11</volume>
<issue>1</issue>
<fpage>16</fpage>
<lpage>16</lpage>
<pub-id pub-id-type="doi">10.1186/s44156-024-00051-2</pub-id>
</element-citation>
<mixed-citation>Robinson S, Ring L, Oxborough D, Harkness A, Bennett S, Rana B, et al. The Assessment of Left Ventricular Diastolic Function: Guidance and Recommendations from the British Society of Echocardiography. Echo Res Pract. 2024;11(1):16. doi: 10.1186/s44156-024-00051-2.</mixed-citation>
</ref>
<ref id="B12">
<label>12</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Lee</surname><given-names>E</given-names></name>
<name><surname>Ito</surname><given-names>S</given-names></name>
<name><surname>Miranda</surname><given-names>WR</given-names></name>
<name><surname>Lopez-Jimenez</surname><given-names>F</given-names></name>
<name><surname>Kane</surname><given-names>GC</given-names></name>
<name><surname>Asirvatham</surname><given-names>SJ</given-names></name>
<etal/>
</person-group>
<article-title>Artificial intelligence-enabled ECG for Left Ventricular Diastolic Function and Filling Pressure</article-title>
<source>NPJ Digit Med</source>
<year>2024</year>
<volume>7</volume>
<issue>1</issue>
<fpage>4</fpage>
<lpage>4</lpage>
<pub-id pub-id-type="doi">10.1038/s41746-023-00993-7</pub-id>
</element-citation>
<mixed-citation>Lee E, Ito S, Miranda WR, Lopez-Jimenez F, Kane GC, Asirvatham SJ, et al. Artificial intelligence-enabled ECG for Left Ventricular Diastolic Function and Filling Pressure. NPJ Digit Med. 2024;7(1):4. doi: 10.1038/s41746-023-00993-7.</mixed-citation>
</ref>
<ref id="B13">
<label>13</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Long</surname><given-names>A</given-names></name>
<name><surname>Haggerty</surname><given-names>CM</given-names></name>
<name><surname>Finer</surname><given-names>J</given-names></name>
<name><surname>Hartzel</surname><given-names>D</given-names></name>
<name><surname>Jing</surname><given-names>L</given-names></name>
<name><surname>Keivani</surname><given-names>A</given-names></name>
<etal/>
</person-group>
<article-title>Deep Learning for Echo Analysis, Tracking, and Evaluation of Mitral Regurgitation (DELINEATE-MR)</article-title>
<source>Circulation</source>
<year>2024</year>
<volume>150</volume>
<issue>12</issue>
<fpage>911</fpage>
<lpage>922</lpage>
<pub-id pub-id-type="doi">10.1161/CIRCULATIONAHA.124.068996</pub-id>
</element-citation>
<mixed-citation>Long A, Haggerty CM, Finer J, Hartzel D, Jing L, Keivani A, et al. Deep Learning for Echo Analysis, Tracking, and Evaluation of Mitral Regurgitation (DELINEATE-MR). Circulation. 2024;150(12):911-22. doi: 10.1161/CIRCULATIONAHA.124.068996.</mixed-citation>
</ref>
<ref id="B14">
<label>14</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Hausleiter</surname><given-names>J</given-names></name>
<name><surname>Lachmann</surname><given-names>M</given-names></name>
<name><surname>Stolz</surname><given-names>L</given-names></name>
<name><surname>Bedogni</surname><given-names>F</given-names></name>
<name><surname>Rubbio</surname><given-names>AP</given-names></name>
<name><surname>Estévez-Loureiro</surname><given-names>R</given-names></name>
<etal/>
</person-group>
<article-title>Artificial Intelligence-derived Risk Score for Mortality in Secondary Mitral Regurgitation Treated by Transcatheter Edge-to-edge Repair: The EuroSMR Risk Score</article-title>
<source>Eur Heart J</source>
<year>2024</year>
<volume>45</volume>
<issue>11</issue>
<fpage>922</fpage>
<lpage>936</lpage>
<pub-id pub-id-type="doi">10.1093/eurheartj/ehad871</pub-id>
</element-citation>
<mixed-citation>Hausleiter J, Lachmann M, Stolz L, Bedogni F, Rubbio AP, Estévez-Loureiro R, et al. Artificial Intelligence-derived Risk Score for Mortality in Secondary Mitral Regurgitation Treated by Transcatheter Edge-to-edge Repair: The EuroSMR Risk Score. Eur Heart J. 2024;45(11):922-36. doi: 10.1093/eurheartj/ehad871.</mixed-citation>
</ref>
<ref id="B15">
<label>15</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Naser</surname><given-names>JA</given-names></name>
<name><surname>Lee</surname><given-names>E</given-names></name>
<name><surname>Pislaru</surname><given-names>SV</given-names></name>
<name><surname>Tsaban</surname><given-names>G</given-names></name>
<name><surname>Malins</surname><given-names>JG</given-names></name>
<name><surname>Jackson</surname><given-names>JI</given-names></name>
<etal/>
</person-group>
<article-title>Artificial Intelligence-based Classification of Echocardiographic Views</article-title>
<source>Eur Heart J Digit Health</source>
<year>2024</year>
<volume>5</volume>
<issue>3</issue>
<fpage>260</fpage>
<lpage>269</lpage>
<pub-id pub-id-type="doi">10.1093/ehjdh/ztae015</pub-id>
</element-citation>
<mixed-citation>Naser JA, Lee E, Pislaru SV, Tsaban G, Malins JG, Jackson JI, et al. Artificial Intelligence-based Classification of Echocardiographic Views. Eur Heart J Digit Health. 2024;5(3):260-9. doi: 10.1093/ehjdh/ztae015.</mixed-citation>
</ref>
<ref id="B16">
<label>16</label>
<element-citation publication-type="journal">
<person-group person-group-type="author">
<name><surname>Kim</surname><given-names>S</given-names></name>
<name><surname>Fischetti</surname><given-names>C</given-names></name>
<name><surname>Guy</surname><given-names>M</given-names></name>
<name><surname>Hsu</surname><given-names>E</given-names></name>
<name><surname>Fox</surname><given-names>J</given-names></name>
<name><surname>Young</surname><given-names>SD</given-names></name>
</person-group>
<article-title>Artificial Intelligence (AI) Applications for Point of Care Ultrasound (POCUS) in Low-resource Settings: A Scoping Review</article-title>
<source>Diagnostics</source>
<year>2024</year>
<volume>14</volume>
<issue>15</issue>
<fpage>1669</fpage>
<lpage>1669</lpage>
<pub-id pub-id-type="doi">10.3390/diagnostics14151669</pub-id>
</element-citation>
<mixed-citation>Kim S, Fischetti C, Guy M, Hsu E, Fox J, Young SD. Artificial Intelligence (AI) Applications for Point of Care Ultrasound (POCUS) in Low-resource Settings: A Scoping Review. Diagnostics. 2024;14(15):1669. doi: 10.3390/diagnostics14151669.</mixed-citation>
</ref>
</ref-list>
</back>
<sub-article article-type="translation" id="S1" xml:lang="pt">
<front-stub>
<article-id pub-id-type="doi">10.36660/abcimg.20240095</article-id>
<article-categories>
<subj-group subj-group-type="heading">
<subject>Editorial</subject>
</subj-group>
</article-categories>
<title-group>
<article-title>Inteligência Artificial em Ecocardiografia: O Futuro do Diagnóstico de Precisão</article-title>
</title-group>
<contrib-group>
<contrib contrib-type="editor">
<contrib-id contrib-id-type="orcid">0000-0002-2634-4632</contrib-id>
<name><surname>Otto</surname><given-names>Maria Estefânia Bosco</given-names></name>
<xref ref-type="aff" rid="aff3"><sup>1</sup></xref>
<xref ref-type="aff" rid="aff4"><sup>2</sup></xref>
<xref ref-type="corresp" rid="c1"/></contrib>
<aff id="aff3">
<label>1</label>
<addr-line>
<named-content content-type="city">Brasília</named-content>
<named-content content-type="state">DF</named-content>
</addr-line>
<country country="BR">Brasil</country>
<institution content-type="original">Instituto de Cardiologia do Distrito Federal (ICDF), Brasília, DF – Brasil</institution>
</aff>
<aff id="aff4">
<label>2</label>
<addr-line>
<named-content content-type="city">Brasília</named-content>
<named-content content-type="state">DF</named-content>
</addr-line>
<country country="BR">Brasil</country>
<institution content-type="original">DF Star, Brasília, DF – Brasil</institution>
</aff>
</contrib-group>
<author-notes>
<corresp id="c2"><label>Correspondência:</label> Maria Estefania Otto • Instituto de Cardiologia do Distrito Federal (ICDF). SQSW 301, Bloco F, Apto 508. CEP: <postal-code>70673-106</postal-code>. Brasília, DF – Brasil. E-mail: <email>mariaestefaniaotto@gmail.com</email></corresp>
</author-notes>
<kwd-group xml:lang="pt">
<title>Palavras-chave</title>
<kwd>Inteligência Artificial</kwd>
<kwd>Ecocardiografia</kwd>
<kwd>Diagnóstico</kwd>
</kwd-group>
</front-stub>
<body>
<p>A integração da inteligência artificial (IA) à ecocardiografia inaugura uma era transformadora na imagem cardiovascular, oferecendo maior precisão diagnóstica, eficiência no fluxo de trabalho e integração de dados. O surgimento de tecnologias baseadas em IA nas diversas áreas da imagem cardiovascular está redefinindo, não somente nossa área de atuação, mas também a interação entre a medicina diagnóstica e a prática clínica. Este editorial tem como objetivo trazer definições básicas sobre termos utilizados na IA resumidos na <xref ref-type="table" rid="t2">Tabela 1</xref><sup><xref ref-type="bibr" rid="B1">1</xref></sup> e explorar o papel crescente da IA na ecocardiografia, com ênfase em suas aplicações na função sistólica e diastólica, nas valvulopatias mitral e aórtica e no crescente uso do ultrassom de <italic>point-of-care</italic> (POCUS). Ademais, faz uma reflexão final das vantagens e desvantagens do uso da IA na medicina diagnóstica.</p>
<table-wrap id="t2">
<label>Tabela 1</label>
<caption><title>Definição de termos associados a IA</title></caption>
<table frame="hsides" rules="groups">
<colgroup width="33%">
<col/>
<col/>
<col/>
</colgroup>
<thead style="border-top: thin solid; border-bottom: thin solid; border-color: #000000">
<tr style="background-color:#C58874">
<th align="left" valign="middle">Termo</th>
<th align="center" valign="middle">Abreviação</th>
<th align="center" valign="middle">Conceito</th>
</tr>
</thead>
<tbody style="border-bottom: thin solid; border-color: #000000">
<tr>
<td align="left" valign="middle">Inteligência Artificial</td>
<td align="center" valign="middle">IA</td>
<td align="center" valign="middle">Tecnologia que permite aos sistemas computacionais, a imitação de funções cognitivas humanas como aprendizado, raciocínio e resolução de problemas.</td>
</tr>
<tr style="background-color:#E8CCBF">
<td align="left" valign="middle">Rede Neural Convolucional</td>
<td align="center" valign="middle">RNC</td>
<td align="center" valign="middle">Arquitetura especializada para analisar imagens visuais, dentro da categoria de redes neurais profundas, utilizando camadas de convolução para capturar padrões espaciais.</td>
</tr>
<tr>
<td align="left" valign="middle">Aprendizado Profundo (deep learning)</td>
<td align="center" valign="middle">AP</td>
<td align="center" valign="middle">Subcampo do AM que usa redes neurais com multicamadas para aprender com grandes conjuntos de dados, eficaz para tarefas complexas, como reconhecimento de imagens.</td>
</tr>
<tr style="background-color:#E8CCBF">
<td align="left" valign="middle">Rede Neural Profunda</td>
<td align="center" valign="middle">RNP</td>
<td align="center" valign="middle">Estrutura de AP projetada para analisar padrões de dados complexos, utilizando várias camadas de processamento para melhorar a precisão de previsões.</td>
</tr>
<tr>
<td align="left" valign="middle">Modelo de Linguagem Grande (large language model)</td>
<td align="center" valign="middle">MLG</td>
<td align="center" valign="middle">Modelos avançados de aprendizado profundo, como o GPT (exatamente o modelo do chat GPT), treinados em grandes volumes de dados textuais, eficazes no processamento e produção de linguagem natural.</td>
</tr>
<tr style="background-color:#E8CCBF">
<td align="left" valign="middle">Aprendizado de Máquina (machine learning)</td>
<td align="center" valign="middle">AM</td>
<td align="center" valign="middle">Ramo da IA que desenvolve algoritmos para treinar computadores a analisar dados e fazer previsões, permitindo aprendizado autônomo a partir de grandes volumes de dados como reconhecer uma incidência de ecocardiograma.</td>
</tr>
<tr>
<td align="left" valign="middle">Processamento de Linguagem Natural</td>
<td align="center" valign="middle">PLN</td>
<td align="center" valign="middle">Campo da IA que capacita máquinas a compreender e interpretar a linguagem humana, crucial para aplicações como tradução de idiomas e análise de sentimentos.</td>
</tr>
</tbody>
</table>
<table-wrap-foot>
<fn id="TFN2"><p>Adaptado de Lüscher et al.<sup><xref ref-type="bibr" rid="B1">1</xref></sup></p></fn>
</table-wrap-foot>
</table-wrap>
<sec>
<title>IA e a Função Sistólica</title>
<p>A fração de ejeção (FE) é um parâmetro essencial para a avaliação da função sistólica. No entanto, sabe-se que historicamente está sujeita à variabilidade interobservador e ao nível de experiência do examinador.<sup><xref ref-type="bibr" rid="B2">2</xref></sup> Algoritmos baseados em IA apresentam soluções para FE, fornecendo cálculos automatizados e reprodutíveis.<sup>3–5</sup> Um estudo pioneiro publicado por He et al.,<sup><xref ref-type="bibr" rid="B6">6</xref></sup> avaliou um sistema de aprendizado profundo (AP) composto por redes neurais convolucionais (RNC), denominado <italic>EchoNet-Dynamic flow</italic> e comparou as avaliações da FE feitas por IA com medidas realizadas por sonógrafo. Os cardiologistas de referência que checaram estes resultados, não conseguiram diferenciar entre sonógrafo e IA. Mais que isso, os resultados demonstraram que a avaliação pela IA não foi inferior às realizadas por especialistas, com uma diferença menor de 5% na variabilidade. Outras vantagens desse sistema foram a rapidez na execução das medidas e a possibilidade de realizar média de até cinco batimentos cardíacos, melhorando a acurácia do resultado na fibrilação atrial.</p>
<p>Outro parâmetro amplamente utilizado para avaliar a função sistólica na ecocardiografia é a análise do Strain Global Longitudinal (SGL),<sup><xref ref-type="bibr" rid="B7">7</xref></sup> que não apenas auxilia no diagnóstico da função cardíaca, mas também revela padrões característicos de certas doenças, como a amiloidose.<sup><xref ref-type="bibr" rid="B7">7</xref></sup> A própria técnica do SGL exige dos equipamentos um treinamento de aprendizado de máquina (AM) para identificação das incidências de janela acústica e delineamento de bordos. No entanto, técnicas de AP por RNC foram capazes de avaliar o SGL mais rapidamente e com menor variabilidade em relação a uma estação de trabalho estabelecida no mercado e utilizada nos laboratórios de ecocardiografia.<sup><xref ref-type="bibr" rid="B8">8</xref></sup> Recentemente, Kwan et al.<sup><xref ref-type="bibr" rid="B9">9</xref></sup> avaliaram um sistema de RNC para avaliação do SGL com imagens de equipamentos de duas empresas distintas, concluindo que, além de ter menor variabilidade e boa reprodutibilidade, esse sistema próprio foi capaz de identificar padrões de SGL e diferenciar entre hipertrofia de atleta, amiloidose e miocardiopatia hipertrófica. No entanto, deve-se considerar que a qualidade da obtenção das imagens tanto para avaliação do SGL como para FE deve ser boa, fator que ainda não pode ser realizado pela IA.</p>
</sec>
<sec>
<title>IA e a Função Diastólica</title>
<p>A avaliação da função diastólica do ventrículo esquerdo desempenha um papel essencial no diagnóstico e prognóstico de diversas doenças cardíacas, principalmente da insuficiência cardíaca com FE preservada. No entanto, os múltiplos algoritmos e a complexidade dos parâmetros ecocardiográficos de avaliação da função diastólica dificultam a classificação adequada em aproximadamente 20% dos pacientes, mesmo com a análise de profissionais experientes.<sup><xref ref-type="bibr" rid="B10">10</xref></sup> Nesse grupo, classificado como função diastólica indeterminada, não é possível definir a presença ou ausência da elevação das pressões de enchimento, parâmetro essencial para determinação de prognóstico e auxílio no tratamento.<sup><xref ref-type="bibr" rid="B11">11</xref></sup></p>
<p>Nesse cenário, dados do eletrocardiograma (ECG), perceptíveis somente pelo AP da IA surgem como uma solução promissora para detectar pacientes com aumento das pressões de enchimento. Lee et al.,<sup><xref ref-type="bibr" rid="B12">12</xref></sup> analisaram mais de 90 mil ECG por IA em pacientes que avaliaram a função diastólica com uma diferença máxima de 14 dias. Ao incorporar o ECG aos dados de ecocardiograma, a acurácia para detecção de pressão de enchimento elevada foi de 91%, rivalizando com as avaliações ecocardiográficas tradicionais. Dessa forma, o uso do ECG avaliado por IA pode complementar os dados do ecocardiograma, especialmente em situações em que o número de parâmetros é muito grande e há falta de dicotomização. A capacidade de prever disfunção diastólica e pressões de enchimento elevadas a partir de um simples ECG sublinha o papel da IA em expandir a capacidade diagnóstica em ambientes com recursos limitados e atualmente foi incorporada ao laudo de ecocardiograma dos autores do artigo.</p>
</sec>
<sec>
<title>IA na Doença Valvar Mitral</title>
<p>A regurgitação mitral (RM) é altamente prevalente na população e está associada a uma significativa redução na qualidade de vida. O ecocardiograma transtorácico (ETT) desempenha um papel essencial para detecção da presença e quantificação da RM. No entanto, está sujeito a variabilidade interobservador significativa, no que tange às medidas de volume e orifício regurgitante. A avaliação de RM pela IA é desafiadora, pois somente algumas imagens obtidas durante o exame contém os dados que definem a gravidade. Mais recentemente, Long et al.<sup><xref ref-type="bibr" rid="B13">13</xref></sup> descreveram um modelo de AP capaz de interpretar dados em vídeo de ETT e classificar automaticamente a gravidade da RM. Esse modelo integra a imagem em vídeo do Doppler colorido, oferecendo uma graduação precisa da RM em uma escala que vai de discreta a acentuada com acurácia em torno de 80%.</p>
<p>Além disso, o poder preditivo da IA é destacado pelo desenvolvimento de ferramentas de estratificação de risco, como o escore <bold>EuroSMR</bold>,<sup><xref ref-type="bibr" rid="B14">14</xref></sup> projetado para prever desfechos em pacientes submetidos à reparação transcateter borda a borda para RM secundária. Essa ferramenta derivada da IA demonstrou desempenho robusto na predição de mortalidade, auxiliando os clínicos na tomada decisões junto ao <italic>Heart Team</italic>. Esse escore pode ser consultado online em <italic><ext-link ext-link-type="uri" xlink:href="http://www.eurosmr.com">www.eurosmr.com</ext-link></italic>.</p>
</sec>
<sec>
<title>IA na Doença Valvar Aórtica</title>
<p>A estenose aórtica (EA) é uma valvopatia muito prevalente e letal, especialmente em pacientes idosos. Novamente, o método de escolha para avaliação da EA é o ETT com a integração de parâmetros de Doppler e a análise da via de saída do ventrículo esquerdo. No entanto, há necessidade de inúmeras variáveis, que além de consumirem tempo para análise, estão sujeitas à variabilidade intra e interobservador. Sistemas de IA totalmente automatizados foram desenvolvidos para enfrentar esses desafios. No estudo de Krishna et al.,<sup><xref ref-type="bibr" rid="B5">5</xref></sup> aplicaram-se algoritmos de AP com RNC a uma coorte de pacientes com diferentes graus de EA. O resultado foi uma excelente correlação com as medidas feitas por cardiologistas experientes, ainda com a vantagem de redução substancial do tempo de análise.</p>
<p>O papel da IA na avaliação da EA não se limita à melhoria da precisão das medidas, mas também se estende à identificação de marcadores precoces da doença. Essas tecnologias são especialmente promissoras em programas de triagem, em que a avaliação rápida e precisa da função valvular é fundamental.</p>
</sec>
<sec>
<title>IA no POCUS</title>
<p>A introdução da IA no POCUS abriu novas possibilidades para avaliações diagnósticas em tempo real, especialmente em ambientes com poucos recursos. Dispositivos de POCUS aprimorados por IA são projetados para auxiliar operadores não especializados, automatizando a aquisição e interpretação de imagens. Por exemplo, a IA pode guiar o operador na obtenção das janelas acústicas corretas, reduzindo significativamente a curva de aprendizado no POCUS.<sup><xref ref-type="bibr" rid="B15">15</xref></sup></p>
<p>As aplicações de IA no POCUS são particularmente benéficas em cenários de emergência, em que a tomada de decisão rápida é essencial. O uso da IA pode facilitar a ultrassonografia pulmonar para detecção de congestão e derrame pleural de forma consistente. Ademais, em áreas rurais e remotas, onde o acesso à interpretação especializada é limitado, a IA pode fornecer suporte diagnóstico em tempo real, democratizando o acesso a tecnologias avançadas.<sup><xref ref-type="bibr" rid="B16">16</xref></sup></p>
</sec>
<sec>
<title>Reflexões sobre a IA na imagem cardiovascular</title>
<p>A IA agiliza e simplifica a avaliação das imagens, otimizando o fluxo de trabalho nos laboratórios de ecocardiografia. No entanto, a aquisição das imagens ainda é realizada manualmente, e, para que a análise por RNC seja precisa, é fundamental que as imagens sejam de alta qualidade, tornando o componente humano indispensável. Além disso, outra limitação da IA é a necessidade de um volume significativo de imagens para treinar adequadamente o modelo de RNC.</p>
<p>Podemos refletir sobre dois pontos de vista opostos. Segundo Yuval Harari, a IA não é uma tecnologia passiva, mas sim um sistema que aprende de forma autônoma, o que pode levar à diminuição do empoderamento humano, conforme descrito em seu livro <italic>Nexus: A Brief History of Information Networks</italic> (2024). Em contrapartida, Kai Fu Lee, um dos principais estudiosos da IA desde o seu início e autor de diversas obras sobre o tema, defende que ‘a IA não substituirá os humanos, mas sim humanos que utilizam IA substituirão aqueles que não a utilizam.</p>
</sec>
<sec sec-type="conclusions">
<title>Conclusão</title>
<p>A IA está prestes a assumir um papel central no futuro da ecocardiografia, elevando a precisão diagnóstica, aumentando a eficiência e otimizando o fluxo de trabalho nos laboratórios de ecocardiograma. No entanto, esses avanços trazem consigo desafios, como a necessidade de validação robusta em populações diversificadas e uma supervisão regulatória rigorosa. À medida que o campo avança, a colaboração entre desenvolvedores de IA, especialistas em imagem cardiovascular e órgãos reguladores será crucial para garantir a integração segura e eficaz da IA à prática médica.</p>
</sec>
</body>

</sub-article>
</article>
