A Deep Learning Pipeline for Assessing Ventricular Volumes from a Cardiac MRI Registry of Patients with Single Ventricle Physiology.

IF 8.1 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE
Tina Yao, Nicole St Clair, Gabriel F Miller, Adam L Dorfman, Mark A Fogel, Sunil Ghelani, Rajesh Krishnamurthy, Christopher Z Lam, Michael Quail, Joshua D Robinson, David Schidlow, Timothy C Slesnick, Justin Weigand, Jennifer A Steeden, Rahul H Rathod, Vivek Muthurangu
{"title":"A Deep Learning Pipeline for Assessing Ventricular Volumes from a Cardiac MRI Registry of Patients with Single Ventricle Physiology.","authors":"Tina Yao, Nicole St Clair, Gabriel F Miller, Adam L Dorfman, Mark A Fogel, Sunil Ghelani, Rajesh Krishnamurthy, Christopher Z Lam, Michael Quail, Joshua D Robinson, David Schidlow, Timothy C Slesnick, Justin Weigand, Jennifer A Steeden, Rahul H Rathod, Vivek Muthurangu","doi":"10.1148/ryai.230132","DOIUrl":null,"url":null,"abstract":"<p><p>Purpose To develop an end-to-end deep learning (DL) pipeline for automated ventricular segmentation of cardiac MRI data from a multicenter registry of patients with Fontan circulation (Fontan Outcomes Registry Using CMR Examinations [FORCE]). Materials and Methods This retrospective study used 250 cardiac MRI examinations (November 2007-December 2022) from 13 institutions for training, validation, and testing. The pipeline contained three DL models: a classifier to identify short-axis cine stacks and two U-Net 3+ models for image cropping and segmentation. The automated segmentations were evaluated on the test set (<i>n</i> = 50) by using the Dice score. Volumetric and functional metrics derived from DL and ground truth manual segmentations were compared using Bland-Altman and intraclass correlation analysis. The pipeline was further qualitatively evaluated on 475 unseen examinations. Results There were acceptable limits of agreement (LOA) and minimal biases between the ground truth and DL end-diastolic volume (EDV) (bias: -0.6 mL/m<sup>2</sup>, LOA: -20.6 to 19.5 mL/m<sup>2</sup>) and end-systolic volume (ESV) (bias: -1.1 mL/m<sup>2</sup>, LOA: -18.1 to 15.9 mL/m<sup>2</sup>), with high intraclass correlation coefficients (ICCs > 0.97) and Dice scores (EDV, 0.91 and ESV, 0.86). There was moderate agreement for ventricular mass (bias: -1.9 g/m<sup>2</sup>, LOA: -17.3 to 13.5 g/m<sup>2</sup>) and an ICC of 0.94. There was also acceptable agreement for stroke volume (bias: 0.6 mL/m<sup>2</sup>, LOA: -17.2 to 18.3 mL/m<sup>2</sup>) and ejection fraction (bias: 0.6%, LOA: -12.2% to 13.4%), with high ICCs (>0.81). The pipeline achieved satisfactory segmentation in 68% of the 475 unseen examinations, while 26% needed minor adjustments, 5% needed major adjustments, and in 0.4%, the cropping model failed. Conclusion The DL pipeline can provide fast standardized segmentation for patients with single ventricle physiology across multiple centers. This pipeline can be applied to all cardiac MRI examinations in the FORCE registry. <b>Keywords:</b> Cardiac, Adults and Pediatrics, MR Imaging, Congenital, Volume Analysis, Segmentation, Quantification <i>Supplemental material is available for this article.</i> © RSNA, 2023.</p>","PeriodicalId":29787,"journal":{"name":"Radiology-Artificial Intelligence","volume":"6 1","pages":"e230132"},"PeriodicalIF":8.1000,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10831511/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Radiology-Artificial Intelligence","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1148/ryai.230132","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
引用次数: 0

Abstract

Purpose To develop an end-to-end deep learning (DL) pipeline for automated ventricular segmentation of cardiac MRI data from a multicenter registry of patients with Fontan circulation (Fontan Outcomes Registry Using CMR Examinations [FORCE]). Materials and Methods This retrospective study used 250 cardiac MRI examinations (November 2007-December 2022) from 13 institutions for training, validation, and testing. The pipeline contained three DL models: a classifier to identify short-axis cine stacks and two U-Net 3+ models for image cropping and segmentation. The automated segmentations were evaluated on the test set (n = 50) by using the Dice score. Volumetric and functional metrics derived from DL and ground truth manual segmentations were compared using Bland-Altman and intraclass correlation analysis. The pipeline was further qualitatively evaluated on 475 unseen examinations. Results There were acceptable limits of agreement (LOA) and minimal biases between the ground truth and DL end-diastolic volume (EDV) (bias: -0.6 mL/m2, LOA: -20.6 to 19.5 mL/m2) and end-systolic volume (ESV) (bias: -1.1 mL/m2, LOA: -18.1 to 15.9 mL/m2), with high intraclass correlation coefficients (ICCs > 0.97) and Dice scores (EDV, 0.91 and ESV, 0.86). There was moderate agreement for ventricular mass (bias: -1.9 g/m2, LOA: -17.3 to 13.5 g/m2) and an ICC of 0.94. There was also acceptable agreement for stroke volume (bias: 0.6 mL/m2, LOA: -17.2 to 18.3 mL/m2) and ejection fraction (bias: 0.6%, LOA: -12.2% to 13.4%), with high ICCs (>0.81). The pipeline achieved satisfactory segmentation in 68% of the 475 unseen examinations, while 26% needed minor adjustments, 5% needed major adjustments, and in 0.4%, the cropping model failed. Conclusion The DL pipeline can provide fast standardized segmentation for patients with single ventricle physiology across multiple centers. This pipeline can be applied to all cardiac MRI examinations in the FORCE registry. Keywords: Cardiac, Adults and Pediatrics, MR Imaging, Congenital, Volume Analysis, Segmentation, Quantification Supplemental material is available for this article. © RSNA, 2023.

从单室生理学患者心脏磁共振成像注册表中评估心室容积的深度学习管道。
目的 开发一种端到端的深度学习(DL)管道,用于对来自丰唐循环患者多中心登记处(Fontan Outcomes Registry Using CMR Examinations [FORCE])的心脏 MRI 数据进行自动心室分割。材料与方法 这项回顾性研究使用了 13 家机构的 250 次心脏 MRI 检查(2007 年 11 月至 2022 年 12 月)进行培训、验证和测试。管道包含三个 DL 模型:一个用于识别短轴电影堆叠的分类器和两个用于图像裁剪和分割的 U-Net 3+ 模型。在测试集(n = 50)上使用 Dice 分数对自动分割进行评估。使用 Bland-Altman 和类内相关分析比较了 DL 和地面实况人工分割得出的体积和功能指标。在 475 例未见检查中对管道进行了进一步的定性评估。结果 地面真实值和 DL 舒张末期容积(EDV)(偏差:-0.6 mL/m2,LOA:-20.6 至 19.5 mL/m2)和收缩末期容积(ESV)(偏差:-1.1 mL/m2,LOA:-18.1 至 15.9 mL/m2)之间存在偏差,具有较高的类内相关系数(ICCs > 0.97)和 Dice 评分(EDV,0.91;ESV,0.86)。心室质量(偏差:-1.9 g/m2,LOA:-17.3 至 13.5 g/m2)的一致性适中,ICC 为 0.94。搏出量(偏差:0.6 mL/m2,LOA:-17.2 至 18.3 mL/m2)和射血分数(偏差:0.6%,LOA:-12.2% 至 13.4%)的一致性也可以接受,ICC 较高(>0.81)。在 475 例未见检查中,该管道有 68% 实现了令人满意的分割,26% 需要微调,5% 需要大幅调整,0.4% 的裁剪模型失败。结论 DL 管道可为多个中心的单心室生理学患者提供快速的标准化分割。该管道可应用于 FORCE 注册中心的所有心脏 MRI 检查。关键词心脏、成人和儿科、磁共振成像、先天性、容积分析、分割、量化 本文有补充材料。© RSNA, 2023.
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
16.20
自引率
1.00%
发文量
0
期刊介绍: Radiology: Artificial Intelligence is a bi-monthly publication that focuses on the emerging applications of machine learning and artificial intelligence in the field of imaging across various disciplines. This journal is available online and accepts multiple manuscript types, including Original Research, Technical Developments, Data Resources, Review articles, Editorials, Letters to the Editor and Replies, Special Reports, and AI in Brief.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信