利用视觉变压器模型提高心脏磁共振对心脏淀粉样变性的诊断准确性。

IF 12.8 1区医学 Q1 CARDIAC & CARDIOVASCULAR SYSTEMS

JACC. Cardiovascular imaging Pub Date : 2025-03-01 DOI:10.1016/j.jcmg.2024.09.010

Joshua Cockrum MD , Makiya Nakashima MS , Carl Ammoury MD , Diane Rizkallah MD , Joseph Mauch MD , David Lopez MD , David Wolinksy MD , Tae Hyun Hwang PhD , Samir Kapadia MD , Lars G. Svensson MD, PhD , Richard Grimm DO , Mazen Hanna MD , W.H. Wilson Tang MD , Christopher Nguyen PhD , David Chen PhD , Deborah Kwon MD

{"title":"利用视觉变压器模型提高心脏磁共振对心脏淀粉样变性的诊断准确性。","authors":"Joshua Cockrum MD , Makiya Nakashima MS , Carl Ammoury MD , Diane Rizkallah MD , Joseph Mauch MD , David Lopez MD , David Wolinksy MD , Tae Hyun Hwang PhD , Samir Kapadia MD , Lars G. Svensson MD, PhD , Richard Grimm DO , Mazen Hanna MD , W.H. Wilson Tang MD , Christopher Nguyen PhD , David Chen PhD , Deborah Kwon MD","doi":"10.1016/j.jcmg.2024.09.010","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>Cardiac magnetic resonance (CMR) imaging is an important diagnostic tool for diagnosis of cardiac amyloidosis (CA). However, discrimination of CA from other etiologies of myocardial disease can be challenging.</div></div><div><h3>Objectives</h3><div>The aim of this study was to develop and rigorously validate a deep learning (DL) algorithm to aid in the discrimination of CA using cine and late gadolinium enhancement CMR imaging.</div></div><div><h3>Methods</h3><div>A DL model using a retrospective cohort of 807 patients who were referred for CMR for suspicion of infiltrative disease or hypertrophic cardiomyopathy (HCM) was developed. Confirmed definitive diagnosis was as follows: 252 patients with CA, 290 patients with HCM, and 265 with neither CA or HCM (other). This cohort was split 70/30 into training and test sets. A vision transformer (ViT) model was trained primarily to identify CA. The model was validated in an external cohort of 157 patients also referred for CMR for suspicion of infiltrative disease or HCM (51 CA, 49 HCM, and 57 other).</div></div><div><h3>Results</h3><div>The ViT model achieved a diagnostic accuracy (84.1%) and an area under the curve of 0.954 in the internal testing data set. The ViT model further demonstrated an accuracy of 82.8% and an area under the curve of 0.957 in the external testing set. The ViT model achieved an accuracy of 90% (n = 55 of 61), among studies with clinical reports with moderate/high confidence diagnosis of CA, and 61.1% (n = 22 of 36) among studies with reported uncertain, missing, or incorrect diagnosis of CA in the internal cohort. DL accuracy of this cohort increased to 79.1% when studies with poor image quality, dual pathologies, or ambiguity of clinically significant CA diagnosis were removed.</div></div><div><h3>Conclusions</h3><div>A ViT model using only cine and late gadolinium enhancement CMR images can achieve high accuracy in differentiating CA from other underlying etiologies of suspected cardiomyopathy, especially in cases when reported human diagnostic confidence was uncertain in both a large single state health system and in an external CA cohort.</div></div>","PeriodicalId":14767,"journal":{"name":"JACC. Cardiovascular imaging","volume":"18 3","pages":"Pages 278-290"},"PeriodicalIF":12.8000,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Leveraging a Vision Transformer Model to Improve Diagnostic Accuracy of Cardiac Amyloidosis With Cardiac Magnetic Resonance\",\"authors\":\"Joshua Cockrum MD , Makiya Nakashima MS , Carl Ammoury MD , Diane Rizkallah MD , Joseph Mauch MD , David Lopez MD , David Wolinksy MD , Tae Hyun Hwang PhD , Samir Kapadia MD , Lars G. Svensson MD, PhD , Richard Grimm DO , Mazen Hanna MD , W.H. Wilson Tang MD , Christopher Nguyen PhD , David Chen PhD , Deborah Kwon MD\",\"doi\":\"10.1016/j.jcmg.2024.09.010\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><h3>Background</h3><div>Cardiac magnetic resonance (CMR) imaging is an important diagnostic tool for diagnosis of cardiac amyloidosis (CA). However, discrimination of CA from other etiologies of myocardial disease can be challenging.</div></div><div><h3>Objectives</h3><div>The aim of this study was to develop and rigorously validate a deep learning (DL) algorithm to aid in the discrimination of CA using cine and late gadolinium enhancement CMR imaging.</div></div><div><h3>Methods</h3><div>A DL model using a retrospective cohort of 807 patients who were referred for CMR for suspicion of infiltrative disease or hypertrophic cardiomyopathy (HCM) was developed. Confirmed definitive diagnosis was as follows: 252 patients with CA, 290 patients with HCM, and 265 with neither CA or HCM (other). This cohort was split 70/30 into training and test sets. A vision transformer (ViT) model was trained primarily to identify CA. The model was validated in an external cohort of 157 patients also referred for CMR for suspicion of infiltrative disease or HCM (51 CA, 49 HCM, and 57 other).</div></div><div><h3>Results</h3><div>The ViT model achieved a diagnostic accuracy (84.1%) and an area under the curve of 0.954 in the internal testing data set. The ViT model further demonstrated an accuracy of 82.8% and an area under the curve of 0.957 in the external testing set. The ViT model achieved an accuracy of 90% (n = 55 of 61), among studies with clinical reports with moderate/high confidence diagnosis of CA, and 61.1% (n = 22 of 36) among studies with reported uncertain, missing, or incorrect diagnosis of CA in the internal cohort. DL accuracy of this cohort increased to 79.1% when studies with poor image quality, dual pathologies, or ambiguity of clinically significant CA diagnosis were removed.</div></div><div><h3>Conclusions</h3><div>A ViT model using only cine and late gadolinium enhancement CMR images can achieve high accuracy in differentiating CA from other underlying etiologies of suspected cardiomyopathy, especially in cases when reported human diagnostic confidence was uncertain in both a large single state health system and in an external CA cohort.</div></div>\",\"PeriodicalId\":14767,\"journal\":{\"name\":\"JACC. Cardiovascular imaging\",\"volume\":\"18 3\",\"pages\":\"Pages 278-290\"},\"PeriodicalIF\":12.8000,\"publicationDate\":\"2025-03-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"JACC. Cardiovascular imaging\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S1936878X24004170\",\"RegionNum\":1,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"CARDIAC & CARDIOVASCULAR SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"JACC. Cardiovascular imaging","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1936878X24004170","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CARDIAC & CARDIOVASCULAR SYSTEMS","Score":null,"Total":0}

引用次数: 0

摘要

背景：心脏磁共振（CMR）成像是诊断心脏淀粉样变性（CA）的重要诊断工具。然而，将CA与其他病因的心肌疾病区分开来是具有挑战性的。目的：本研究的目的是开发并严格验证一种深度学习（DL）算法，以帮助使用电影和晚期钆增强CMR成像来区分CA。方法：对807例因怀疑浸润性疾病或肥厚性心肌病（HCM）而行CMR的患者进行回顾性队列研究，建立DL模型。确诊的最终诊断如下：252例CA， 290例HCM， 265例CA和HCM均无（其他）。该队列按70/30分成训练集和测试集。视觉变压器（ViT）模型主要用于识别CA。该模型在157名因怀疑浸润性疾病或HCM而进行CMR的患者（51名CA， 49名HCM， 57名其他）的外部队列中得到验证。结果：ViT模型在内测数据集中的诊断准确率为84.1%，曲线下面积为0.954。在外部测试集中，ViT模型的准确率为82.8%，曲线下面积为0.957。在临床报告中CA诊断为中等/高置信度的研究中，ViT模型的准确率为90% (n = 55 / 61)，在内部队列中CA诊断不确定、缺失或错误的研究中，ViT模型的准确率为61.1% （n = 22 / 36）。当去除图像质量差、双重病理或临床意义不明确的CA诊断时，该队列的DL准确率增加到79.1%。结论：仅使用电影和晚期钆增强CMR图像的ViT模型可以在区分CA与疑似心肌病的其他潜在病因方面达到很高的准确性，特别是在大型单一状态卫生系统和外部CA队列中报告的人类诊断信心不确定的情况下。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Leveraging a Vision Transformer Model to Improve Diagnostic Accuracy of Cardiac Amyloidosis With Cardiac Magnetic Resonance

Background

Cardiac magnetic resonance (CMR) imaging is an important diagnostic tool for diagnosis of cardiac amyloidosis (CA). However, discrimination of CA from other etiologies of myocardial disease can be challenging.

Objectives

The aim of this study was to develop and rigorously validate a deep learning (DL) algorithm to aid in the discrimination of CA using cine and late gadolinium enhancement CMR imaging.

Methods

A DL model using a retrospective cohort of 807 patients who were referred for CMR for suspicion of infiltrative disease or hypertrophic cardiomyopathy (HCM) was developed. Confirmed definitive diagnosis was as follows: 252 patients with CA, 290 patients with HCM, and 265 with neither CA or HCM (other). This cohort was split 70/30 into training and test sets. A vision transformer (ViT) model was trained primarily to identify CA. The model was validated in an external cohort of 157 patients also referred for CMR for suspicion of infiltrative disease or HCM (51 CA, 49 HCM, and 57 other).

Results

The ViT model achieved a diagnostic accuracy (84.1%) and an area under the curve of 0.954 in the internal testing data set. The ViT model further demonstrated an accuracy of 82.8% and an area under the curve of 0.957 in the external testing set. The ViT model achieved an accuracy of 90% (n = 55 of 61), among studies with clinical reports with moderate/high confidence diagnosis of CA, and 61.1% (n = 22 of 36) among studies with reported uncertain, missing, or incorrect diagnosis of CA in the internal cohort. DL accuracy of this cohort increased to 79.1% when studies with poor image quality, dual pathologies, or ambiguity of clinically significant CA diagnosis were removed.

Conclusions

A ViT model using only cine and late gadolinium enhancement CMR images can achieve high accuracy in differentiating CA from other underlying etiologies of suspected cardiomyopathy, especially in cases when reported human diagnostic confidence was uncertain in both a large single state health system and in an external CA cohort.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

JACC. Cardiovascular imaging CARDIAC & CARDIOVASCULAR SYSTEMS-RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

CiteScore

24.90

自引率

5.70%

发文量

330

审稿时长

4-8 weeks

期刊介绍： JACC: Cardiovascular Imaging, part of the prestigious Journal of the American College of Cardiology (JACC) family, offers readers a comprehensive perspective on all aspects of cardiovascular imaging. This specialist journal covers original clinical research on both non-invasive and invasive imaging techniques, including echocardiography, CT, CMR, nuclear, optical imaging, and cine-angiography. JACC. Cardiovascular imaging highlights advances in basic science and molecular imaging that are expected to significantly impact clinical practice in the next decade. This influence encompasses improvements in diagnostic performance, enhanced understanding of the pathogenetic basis of diseases, and advancements in therapy. In addition to cutting-edge research,the content of JACC: Cardiovascular Imaging emphasizes practical aspects for the practicing cardiologist, including advocacy and practice management.The journal also features state-of-the-art reviews, ensuring a well-rounded and insightful resource for professionals in the field of cardiovascular imaging.