Automated cardiovascular magnetic resonance image analysis with fully convolutional networks 08 Information and Computing Sciences 0801 Artificial Intelligence and Image Processing

Wenjia Bai, Matthew Sinclair, Giacomo Tarroni, Ozan Oktay, Martin Rajchl, Ghislain Vaillant, Aaron M. Lee, Nay Aung, Elena Lukaschuk, Mihir M. Sanghvi, Filip Zemrak, Kenneth Fung, Jose Miguel Paiva, Valentina Carapella, Young Jin Kim, Hideaki Suzuki, Bernhard Kainz, Paul M. Matthews, Steffen E. Petersen, Stefan K. PiechnikStefan Neubauer, Ben Glocker, Daniel Rueckert

Research output: Contribution to journalArticle

47 Citations (Scopus)

Abstract

Background: Cardiovascular resonance (CMR) imaging is a standard imaging modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification of the cardiac chamber volume, ejection fraction and myocardial mass, providing information for diagnosis and monitoring of CVDs. However, for years, clinicians have been relying on manual approaches for CMR image analysis, which is time consuming and prone to subjective errors. It is a major clinical challenge to automatically derive quantitative and clinically relevant information from CMR images. Methods: Deep neural networks have shown a great potential in image pattern recognition and segmentation for a variety of tasks. Here we demonstrate an automated analysis method for CMR images, which is based on a fully convolutional network (FCN). The network is trained and evaluated on a large-scale dataset from the UK Biobank, consisting of 4,875 subjects with 93,500 pixelwise annotated images. The performance of the method has been evaluated using a number of technical metrics, including the Dice metric, mean contour distance and Hausdorff distance, as well as clinically relevant measures, including left ventricle (LV) end-diastolic volume (LVEDV) and end-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolic volume (RVEDV) and end-systolic volume (RVESV). Results: By combining FCN with a large-scale annotated dataset, the proposed automated method achieves a high performance in segmenting the LV and RV on short-axis CMR images and the left atrium (LA) and right atrium (RA) on long-axis CMR images. On a short-axis image test set of 600 subjects, it achieves an average Dice metric of 0.94 for the LV cavity, 0.88 for the LV myocardium and 0.90 for the RV cavity. The mean absolute difference between automated measurement and manual measurement is 6.1 mL for LVEDV, 5.3 mL for LVESV, 6.9 gram for LVM, 8.5 mL for RVEDV and 7.2 mL for RVESV. On long-axis image test sets, the average Dice metric is 0.93 for the LA cavity (2-chamber view), 0.95 for the LA cavity (4-chamber view) and 0.96 for the RA cavity (4-chamber view). The performance is comparable to human inter-observer variability. Conclusions: We show that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinically relevant measures.

Original languageEnglish
Article number65
JournalJournal of Cardiovascular Magnetic Resonance
Volume20
Issue number1
DOIs
Publication statusPublished - 2018 Sep 14

Fingerprint

Information Science
Artificial Intelligence
Heart Ventricles
Magnetic Resonance Spectroscopy
Heart Atria
Cardiovascular Diseases
Cardiac Volume
Observer Variation
Cause of Death
Myocardium

All Science Journal Classification (ASJC) codes

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging
  • Cardiology and Cardiovascular Medicine
  • Family Practice

Cite this

Bai, Wenjia ; Sinclair, Matthew ; Tarroni, Giacomo ; Oktay, Ozan ; Rajchl, Martin ; Vaillant, Ghislain ; Lee, Aaron M. ; Aung, Nay ; Lukaschuk, Elena ; Sanghvi, Mihir M. ; Zemrak, Filip ; Fung, Kenneth ; Paiva, Jose Miguel ; Carapella, Valentina ; Kim, Young Jin ; Suzuki, Hideaki ; Kainz, Bernhard ; Matthews, Paul M. ; Petersen, Steffen E. ; Piechnik, Stefan K. ; Neubauer, Stefan ; Glocker, Ben ; Rueckert, Daniel. / Automated cardiovascular magnetic resonance image analysis with fully convolutional networks 08 Information and Computing Sciences 0801 Artificial Intelligence and Image Processing. In: Journal of Cardiovascular Magnetic Resonance. 2018 ; Vol. 20, No. 1.
@article{7b6b6f2641b74b5abcd1cdd0954be974,
title = "Automated cardiovascular magnetic resonance image analysis with fully convolutional networks 08 Information and Computing Sciences 0801 Artificial Intelligence and Image Processing",
abstract = "Background: Cardiovascular resonance (CMR) imaging is a standard imaging modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification of the cardiac chamber volume, ejection fraction and myocardial mass, providing information for diagnosis and monitoring of CVDs. However, for years, clinicians have been relying on manual approaches for CMR image analysis, which is time consuming and prone to subjective errors. It is a major clinical challenge to automatically derive quantitative and clinically relevant information from CMR images. Methods: Deep neural networks have shown a great potential in image pattern recognition and segmentation for a variety of tasks. Here we demonstrate an automated analysis method for CMR images, which is based on a fully convolutional network (FCN). The network is trained and evaluated on a large-scale dataset from the UK Biobank, consisting of 4,875 subjects with 93,500 pixelwise annotated images. The performance of the method has been evaluated using a number of technical metrics, including the Dice metric, mean contour distance and Hausdorff distance, as well as clinically relevant measures, including left ventricle (LV) end-diastolic volume (LVEDV) and end-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolic volume (RVEDV) and end-systolic volume (RVESV). Results: By combining FCN with a large-scale annotated dataset, the proposed automated method achieves a high performance in segmenting the LV and RV on short-axis CMR images and the left atrium (LA) and right atrium (RA) on long-axis CMR images. On a short-axis image test set of 600 subjects, it achieves an average Dice metric of 0.94 for the LV cavity, 0.88 for the LV myocardium and 0.90 for the RV cavity. The mean absolute difference between automated measurement and manual measurement is 6.1 mL for LVEDV, 5.3 mL for LVESV, 6.9 gram for LVM, 8.5 mL for RVEDV and 7.2 mL for RVESV. On long-axis image test sets, the average Dice metric is 0.93 for the LA cavity (2-chamber view), 0.95 for the LA cavity (4-chamber view) and 0.96 for the RA cavity (4-chamber view). The performance is comparable to human inter-observer variability. Conclusions: We show that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinically relevant measures.",
author = "Wenjia Bai and Matthew Sinclair and Giacomo Tarroni and Ozan Oktay and Martin Rajchl and Ghislain Vaillant and Lee, {Aaron M.} and Nay Aung and Elena Lukaschuk and Sanghvi, {Mihir M.} and Filip Zemrak and Kenneth Fung and Paiva, {Jose Miguel} and Valentina Carapella and Kim, {Young Jin} and Hideaki Suzuki and Bernhard Kainz and Matthews, {Paul M.} and Petersen, {Steffen E.} and Piechnik, {Stefan K.} and Stefan Neubauer and Ben Glocker and Daniel Rueckert",
year = "2018",
month = "9",
day = "14",
doi = "10.1186/s12968-018-0471-x",
language = "English",
volume = "20",
journal = "Journal of Cardiovascular Magnetic Resonance",
issn = "1097-6647",
publisher = "BioMed Central",
number = "1",

}

Bai, W, Sinclair, M, Tarroni, G, Oktay, O, Rajchl, M, Vaillant, G, Lee, AM, Aung, N, Lukaschuk, E, Sanghvi, MM, Zemrak, F, Fung, K, Paiva, JM, Carapella, V, Kim, YJ, Suzuki, H, Kainz, B, Matthews, PM, Petersen, SE, Piechnik, SK, Neubauer, S, Glocker, B & Rueckert, D 2018, 'Automated cardiovascular magnetic resonance image analysis with fully convolutional networks 08 Information and Computing Sciences 0801 Artificial Intelligence and Image Processing', Journal of Cardiovascular Magnetic Resonance, vol. 20, no. 1, 65. https://doi.org/10.1186/s12968-018-0471-x

Automated cardiovascular magnetic resonance image analysis with fully convolutional networks 08 Information and Computing Sciences 0801 Artificial Intelligence and Image Processing. / Bai, Wenjia; Sinclair, Matthew; Tarroni, Giacomo; Oktay, Ozan; Rajchl, Martin; Vaillant, Ghislain; Lee, Aaron M.; Aung, Nay; Lukaschuk, Elena; Sanghvi, Mihir M.; Zemrak, Filip; Fung, Kenneth; Paiva, Jose Miguel; Carapella, Valentina; Kim, Young Jin; Suzuki, Hideaki; Kainz, Bernhard; Matthews, Paul M.; Petersen, Steffen E.; Piechnik, Stefan K.; Neubauer, Stefan; Glocker, Ben; Rueckert, Daniel.

In: Journal of Cardiovascular Magnetic Resonance, Vol. 20, No. 1, 65, 14.09.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Automated cardiovascular magnetic resonance image analysis with fully convolutional networks 08 Information and Computing Sciences 0801 Artificial Intelligence and Image Processing

AU - Bai, Wenjia

AU - Sinclair, Matthew

AU - Tarroni, Giacomo

AU - Oktay, Ozan

AU - Rajchl, Martin

AU - Vaillant, Ghislain

AU - Lee, Aaron M.

AU - Aung, Nay

AU - Lukaschuk, Elena

AU - Sanghvi, Mihir M.

AU - Zemrak, Filip

AU - Fung, Kenneth

AU - Paiva, Jose Miguel

AU - Carapella, Valentina

AU - Kim, Young Jin

AU - Suzuki, Hideaki

AU - Kainz, Bernhard

AU - Matthews, Paul M.

AU - Petersen, Steffen E.

AU - Piechnik, Stefan K.

AU - Neubauer, Stefan

AU - Glocker, Ben

AU - Rueckert, Daniel

PY - 2018/9/14

Y1 - 2018/9/14

N2 - Background: Cardiovascular resonance (CMR) imaging is a standard imaging modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification of the cardiac chamber volume, ejection fraction and myocardial mass, providing information for diagnosis and monitoring of CVDs. However, for years, clinicians have been relying on manual approaches for CMR image analysis, which is time consuming and prone to subjective errors. It is a major clinical challenge to automatically derive quantitative and clinically relevant information from CMR images. Methods: Deep neural networks have shown a great potential in image pattern recognition and segmentation for a variety of tasks. Here we demonstrate an automated analysis method for CMR images, which is based on a fully convolutional network (FCN). The network is trained and evaluated on a large-scale dataset from the UK Biobank, consisting of 4,875 subjects with 93,500 pixelwise annotated images. The performance of the method has been evaluated using a number of technical metrics, including the Dice metric, mean contour distance and Hausdorff distance, as well as clinically relevant measures, including left ventricle (LV) end-diastolic volume (LVEDV) and end-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolic volume (RVEDV) and end-systolic volume (RVESV). Results: By combining FCN with a large-scale annotated dataset, the proposed automated method achieves a high performance in segmenting the LV and RV on short-axis CMR images and the left atrium (LA) and right atrium (RA) on long-axis CMR images. On a short-axis image test set of 600 subjects, it achieves an average Dice metric of 0.94 for the LV cavity, 0.88 for the LV myocardium and 0.90 for the RV cavity. The mean absolute difference between automated measurement and manual measurement is 6.1 mL for LVEDV, 5.3 mL for LVESV, 6.9 gram for LVM, 8.5 mL for RVEDV and 7.2 mL for RVESV. On long-axis image test sets, the average Dice metric is 0.93 for the LA cavity (2-chamber view), 0.95 for the LA cavity (4-chamber view) and 0.96 for the RA cavity (4-chamber view). The performance is comparable to human inter-observer variability. Conclusions: We show that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinically relevant measures.

AB - Background: Cardiovascular resonance (CMR) imaging is a standard imaging modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification of the cardiac chamber volume, ejection fraction and myocardial mass, providing information for diagnosis and monitoring of CVDs. However, for years, clinicians have been relying on manual approaches for CMR image analysis, which is time consuming and prone to subjective errors. It is a major clinical challenge to automatically derive quantitative and clinically relevant information from CMR images. Methods: Deep neural networks have shown a great potential in image pattern recognition and segmentation for a variety of tasks. Here we demonstrate an automated analysis method for CMR images, which is based on a fully convolutional network (FCN). The network is trained and evaluated on a large-scale dataset from the UK Biobank, consisting of 4,875 subjects with 93,500 pixelwise annotated images. The performance of the method has been evaluated using a number of technical metrics, including the Dice metric, mean contour distance and Hausdorff distance, as well as clinically relevant measures, including left ventricle (LV) end-diastolic volume (LVEDV) and end-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolic volume (RVEDV) and end-systolic volume (RVESV). Results: By combining FCN with a large-scale annotated dataset, the proposed automated method achieves a high performance in segmenting the LV and RV on short-axis CMR images and the left atrium (LA) and right atrium (RA) on long-axis CMR images. On a short-axis image test set of 600 subjects, it achieves an average Dice metric of 0.94 for the LV cavity, 0.88 for the LV myocardium and 0.90 for the RV cavity. The mean absolute difference between automated measurement and manual measurement is 6.1 mL for LVEDV, 5.3 mL for LVESV, 6.9 gram for LVM, 8.5 mL for RVEDV and 7.2 mL for RVESV. On long-axis image test sets, the average Dice metric is 0.93 for the LA cavity (2-chamber view), 0.95 for the LA cavity (4-chamber view) and 0.96 for the RA cavity (4-chamber view). The performance is comparable to human inter-observer variability. Conclusions: We show that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinically relevant measures.

UR - http://www.scopus.com/inward/record.url?scp=85053360253&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85053360253&partnerID=8YFLogxK

U2 - 10.1186/s12968-018-0471-x

DO - 10.1186/s12968-018-0471-x

M3 - Article

C2 - 30217194

AN - SCOPUS:85053360253

VL - 20

JO - Journal of Cardiovascular Magnetic Resonance

JF - Journal of Cardiovascular Magnetic Resonance

SN - 1097-6647

IS - 1

M1 - 65

ER -