Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning

Lohendran Baskaran; Subhi J. Al’Aref; Gabriel Maliakal; Benjamin C. Lee; Zhuoran Xu; Jeong W. Choi; Sang Eun Lee; Ji Min Sung; Fay Y. Lin; Simon Dunham; Bobak Mosadegh; Yong Jin Kim; Ilan Gottlieb; Byoung Kwon Lee; Eun Ju Chun; Filippo Cademartiri; Erica Maffei; Hugo Marques; Sanghoon Shin; Jung Hyun Choi; Kavitha Chinnaiyan; Martin Hadamitzky; Edoardo Conte; Daniele Andreini; Gianluca Pontone; Matthew J. Budoff; Jonathon A. Leipsic; Gilbert L. Raff; Renu Virmani; Habib Samady; Peter H. Stone; Daniel S. Berman; Jagat Narula; Jeroen J. Bax; Hyuk Jae Chang; James K. Min; Leslee J. Shaw

doi:10.1371/journal.pone.0232573

Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning

Lohendran Baskaran, Subhi J. Al’Aref, Gabriel Maliakal, Benjamin C. Lee, Zhuoran Xu, Jeong W. Choi, Sang Eun Lee, Ji Min Sung, Fay Y. Lin, Simon Dunham, Bobak Mosadegh, Yong Jin Kim, Ilan Gottlieb, Byoung Kwon Lee, Eun Ju Chun, Filippo Cademartiri, Erica Maffei, Hugo Marques, Sanghoon Shin, Jung Hyun ChoiKavitha Chinnaiyan, Martin Hadamitzky, Edoardo Conte, Daniele Andreini, Gianluca Pontone, Matthew J. Budoff, Jonathon A. Leipsic, Gilbert L. Raff, Renu Virmani, Habib Samady, Peter H. Stone, Daniel S. Berman, Jagat Narula, Jeroen J. Bax, Hyuk Jae Chang, James K. Min, Leslee J. Shaw

Department of Internal Medicine

Research output: Contribution to journal › Article › peer-review

12 Citations (Scopus)

Abstract

Objectives To develop, demonstrate and evaluate an automated deep learning method for multiple cardiovascular structure segmentation. Background Segmentation of cardiovascular images is resource-intensive. We design an automated deep learning method for the segmentation of multiple structures from Coronary Computed Tomography Angiography (CCTA) images. Methods Images from a multicenter registry of patients that underwent clinically-indicated CCTA were used. The proximal ascending and descending aorta (PAA, DA), superior and inferior vena cavae (SVC, IVC), pulmonary artery (PA), coronary sinus (CS), right ventricular wall (RVW) and left atrial wall (LAW) were annotated as ground truth. The U-net-derived deep learning model was trained, validated and tested in a 70:20:10 split. Results The dataset comprised 206 patients, with 5.130 billion pixels. Mean age was 59.9 ± 9.4 yrs., and was 42.7% female. An overall median Dice score of 0.820 (0.782, 0.843) was achieved. Median Dice scores for PAA, DA, SVC, IVC, PA, CS, RVW and LAW were 0.969 (0.979, 0.988), 0.953 (0.955, 0.983), 0.937 (0.934, 0.965), 0.903 (0.897, 0.948), 0.775 (0.724, 0.925), 0.720 (0.642, 0.809), 0.685 (0.631, 0.761) and 0.625 (0.596, 0.749) respectively. Apart from the CS, there were no significant differences in performance between sexes or age groups. Conclusions An automated deep learning model demonstrated segmentation of multiple cardiovascular structures from CCTA images with reasonable overall accuracy when evaluated on a pixel level.

Original language	English
Article number	e0232573
Journal	PloS one
Volume	15
Issue number	5
DOIs	https://doi.org/10.1371/journal.pone.0232573
Publication status	Published - 2020 May

Bibliographical note

Funding Information:
Funding: The research reported in this manuscript was supported by the Dalio Institute of Cardiovascular Imaging (New York, NY, USA). James K. Min received funding from the Dalio Foundation, National Institutes of Health, and GE Healthcare. Dr. Min serves on the scientific advisory board of Arineta and GE Healthcare, and became founder and an employee of Cleerly, Inc after this research was conducted. Dr. Jonathon Leipsic is a consultant and has reported stock options with Circle CVI and HeartFlow. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section.

Publisher Copyright:
© 2020 Baskaran et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

All Science Journal Classification (ASJC) codes

General

Access to Document

10.1371/journal.pone.0232573

Cite this

Baskaran, L., Al’Aref, S. J., Maliakal, G., Lee, B. C., Xu, Z., Choi, J. W., Lee, S. E., Sung, J. M., Lin, F. Y., Dunham, S., Mosadegh, B., Kim, Y. J., Gottlieb, I., Lee, B. K., Chun, E. J., Cademartiri, F., Maffei, E., Marques, H., Shin, S., ... Shaw, L. J. (2020). Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning. PloS one, 15(5), [e0232573]. https://doi.org/10.1371/journal.pone.0232573

@article{7822e27e1b894f998592a83c037a3fa3,

title = "Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning",

abstract = "Objectives To develop, demonstrate and evaluate an automated deep learning method for multiple cardiovascular structure segmentation. Background Segmentation of cardiovascular images is resource-intensive. We design an automated deep learning method for the segmentation of multiple structures from Coronary Computed Tomography Angiography (CCTA) images. Methods Images from a multicenter registry of patients that underwent clinically-indicated CCTA were used. The proximal ascending and descending aorta (PAA, DA), superior and inferior vena cavae (SVC, IVC), pulmonary artery (PA), coronary sinus (CS), right ventricular wall (RVW) and left atrial wall (LAW) were annotated as ground truth. The U-net-derived deep learning model was trained, validated and tested in a 70:20:10 split. Results The dataset comprised 206 patients, with 5.130 billion pixels. Mean age was 59.9 ± 9.4 yrs., and was 42.7% female. An overall median Dice score of 0.820 (0.782, 0.843) was achieved. Median Dice scores for PAA, DA, SVC, IVC, PA, CS, RVW and LAW were 0.969 (0.979, 0.988), 0.953 (0.955, 0.983), 0.937 (0.934, 0.965), 0.903 (0.897, 0.948), 0.775 (0.724, 0.925), 0.720 (0.642, 0.809), 0.685 (0.631, 0.761) and 0.625 (0.596, 0.749) respectively. Apart from the CS, there were no significant differences in performance between sexes or age groups. Conclusions An automated deep learning model demonstrated segmentation of multiple cardiovascular structures from CCTA images with reasonable overall accuracy when evaluated on a pixel level.",

author = "Lohendran Baskaran and Al{\textquoteright}Aref, {Subhi J.} and Gabriel Maliakal and Lee, {Benjamin C.} and Zhuoran Xu and Choi, {Jeong W.} and Lee, {Sang Eun} and Sung, {Ji Min} and Lin, {Fay Y.} and Simon Dunham and Bobak Mosadegh and Kim, {Yong Jin} and Ilan Gottlieb and Lee, {Byoung Kwon} and Chun, {Eun Ju} and Filippo Cademartiri and Erica Maffei and Hugo Marques and Sanghoon Shin and Choi, {Jung Hyun} and Kavitha Chinnaiyan and Martin Hadamitzky and Edoardo Conte and Daniele Andreini and Gianluca Pontone and Budoff, {Matthew J.} and Leipsic, {Jonathon A.} and Raff, {Gilbert L.} and Renu Virmani and Habib Samady and Stone, {Peter H.} and Berman, {Daniel S.} and Jagat Narula and Bax, {Jeroen J.} and Chang, {Hyuk Jae} and Min, {James K.} and Shaw, {Leslee J.}",

note = "Funding Information: Funding: The research reported in this manuscript was supported by the Dalio Institute of Cardiovascular Imaging (New York, NY, USA). James K. Min received funding from the Dalio Foundation, National Institutes of Health, and GE Healthcare. Dr. Min serves on the scientific advisory board of Arineta and GE Healthcare, and became founder and an employee of Cleerly, Inc after this research was conducted. Dr. Jonathon Leipsic is a consultant and has reported stock options with Circle CVI and HeartFlow. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the {\textquoteleft}author contributions{\textquoteright} section. Publisher Copyright: {\textcopyright} 2020 Baskaran et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.",

year = "2020",

month = may,

doi = "10.1371/journal.pone.0232573",

language = "English",

volume = "15",

journal = "PLoS One",

issn = "1932-6203",

publisher = "Public Library of Science",

number = "5",

}

Baskaran, L, Al’Aref, SJ, Maliakal, G, Lee, BC, Xu, Z, Choi, JW, Lee, SE, Sung, JM, Lin, FY, Dunham, S, Mosadegh, B, Kim, YJ, Gottlieb, I, Lee, BK, Chun, EJ, Cademartiri, F, Maffei, E, Marques, H, Shin, S, Choi, JH, Chinnaiyan, K, Hadamitzky, M, Conte, E, Andreini, D, Pontone, G, Budoff, MJ, Leipsic, JA, Raff, GL, Virmani, R, Samady, H, Stone, PH, Berman, DS, Narula, J, Bax, JJ, Chang, HJ, Min, JK & Shaw, LJ 2020, 'Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning', PloS one, vol. 15, no. 5, e0232573. https://doi.org/10.1371/journal.pone.0232573

TY - JOUR

T1 - Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning

AU - Baskaran, Lohendran

AU - Al’Aref, Subhi J.

AU - Maliakal, Gabriel

AU - Lee, Benjamin C.

AU - Xu, Zhuoran

AU - Choi, Jeong W.

AU - Lee, Sang Eun

AU - Sung, Ji Min

AU - Lin, Fay Y.

AU - Dunham, Simon

AU - Mosadegh, Bobak

AU - Kim, Yong Jin

AU - Gottlieb, Ilan

AU - Lee, Byoung Kwon

AU - Chun, Eun Ju

AU - Cademartiri, Filippo

AU - Maffei, Erica

AU - Marques, Hugo

AU - Shin, Sanghoon

AU - Choi, Jung Hyun

AU - Chinnaiyan, Kavitha

AU - Hadamitzky, Martin

AU - Conte, Edoardo

AU - Andreini, Daniele

AU - Pontone, Gianluca

AU - Budoff, Matthew J.

AU - Leipsic, Jonathon A.

AU - Raff, Gilbert L.

AU - Virmani, Renu

AU - Samady, Habib

AU - Stone, Peter H.

AU - Berman, Daniel S.

AU - Narula, Jagat

AU - Bax, Jeroen J.

AU - Chang, Hyuk Jae

AU - Min, James K.

AU - Shaw, Leslee J.

N1 - Funding Information: Funding: The research reported in this manuscript was supported by the Dalio Institute of Cardiovascular Imaging (New York, NY, USA). James K. Min received funding from the Dalio Foundation, National Institutes of Health, and GE Healthcare. Dr. Min serves on the scientific advisory board of Arineta and GE Healthcare, and became founder and an employee of Cleerly, Inc after this research was conducted. Dr. Jonathon Leipsic is a consultant and has reported stock options with Circle CVI and HeartFlow. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section. Publisher Copyright: © 2020 Baskaran et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

PY - 2020/5

Y1 - 2020/5

N2 - Objectives To develop, demonstrate and evaluate an automated deep learning method for multiple cardiovascular structure segmentation. Background Segmentation of cardiovascular images is resource-intensive. We design an automated deep learning method for the segmentation of multiple structures from Coronary Computed Tomography Angiography (CCTA) images. Methods Images from a multicenter registry of patients that underwent clinically-indicated CCTA were used. The proximal ascending and descending aorta (PAA, DA), superior and inferior vena cavae (SVC, IVC), pulmonary artery (PA), coronary sinus (CS), right ventricular wall (RVW) and left atrial wall (LAW) were annotated as ground truth. The U-net-derived deep learning model was trained, validated and tested in a 70:20:10 split. Results The dataset comprised 206 patients, with 5.130 billion pixels. Mean age was 59.9 ± 9.4 yrs., and was 42.7% female. An overall median Dice score of 0.820 (0.782, 0.843) was achieved. Median Dice scores for PAA, DA, SVC, IVC, PA, CS, RVW and LAW were 0.969 (0.979, 0.988), 0.953 (0.955, 0.983), 0.937 (0.934, 0.965), 0.903 (0.897, 0.948), 0.775 (0.724, 0.925), 0.720 (0.642, 0.809), 0.685 (0.631, 0.761) and 0.625 (0.596, 0.749) respectively. Apart from the CS, there were no significant differences in performance between sexes or age groups. Conclusions An automated deep learning model demonstrated segmentation of multiple cardiovascular structures from CCTA images with reasonable overall accuracy when evaluated on a pixel level.

AB - Objectives To develop, demonstrate and evaluate an automated deep learning method for multiple cardiovascular structure segmentation. Background Segmentation of cardiovascular images is resource-intensive. We design an automated deep learning method for the segmentation of multiple structures from Coronary Computed Tomography Angiography (CCTA) images. Methods Images from a multicenter registry of patients that underwent clinically-indicated CCTA were used. The proximal ascending and descending aorta (PAA, DA), superior and inferior vena cavae (SVC, IVC), pulmonary artery (PA), coronary sinus (CS), right ventricular wall (RVW) and left atrial wall (LAW) were annotated as ground truth. The U-net-derived deep learning model was trained, validated and tested in a 70:20:10 split. Results The dataset comprised 206 patients, with 5.130 billion pixels. Mean age was 59.9 ± 9.4 yrs., and was 42.7% female. An overall median Dice score of 0.820 (0.782, 0.843) was achieved. Median Dice scores for PAA, DA, SVC, IVC, PA, CS, RVW and LAW were 0.969 (0.979, 0.988), 0.953 (0.955, 0.983), 0.937 (0.934, 0.965), 0.903 (0.897, 0.948), 0.775 (0.724, 0.925), 0.720 (0.642, 0.809), 0.685 (0.631, 0.761) and 0.625 (0.596, 0.749) respectively. Apart from the CS, there were no significant differences in performance between sexes or age groups. Conclusions An automated deep learning model demonstrated segmentation of multiple cardiovascular structures from CCTA images with reasonable overall accuracy when evaluated on a pixel level.

UR - http://www.scopus.com/inward/record.url?scp=85084406001&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85084406001&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0232573

DO - 10.1371/journal.pone.0232573

M3 - Article

C2 - 32374784

AN - SCOPUS:85084406001

SN - 1932-6203

VL - 15

JO - PLoS One

JF - PLoS One

IS - 5

M1 - e0232573

ER -

Automatic segmentation of multiple cardiovascular structures from cardiac computed tomography angiography images using deep learning

Abstract

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this