Cardiac MRI Segmentation of Ventricular Structures and Myocardium Using U-Net Variants

Chinthalanka Wijesinghe; Dulani Meedeniya; Pratheepan Yogarajah

doi:10.1109/ICARC64760.2025.10963166

Cardiac MRI Segmentation of Ventricular Structures and Myocardium Using U-Net Variants

Chinthalanka Wijesinghe
, Dulani Meedeniya
, Pratheepan Yogarajah

University of Moratuwa

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Citations (Scopus)

141 Downloads (Pure)

Abstract

Accurate segmentation of the ventricular structures and myocardium from Cardiac Magnetic Resonance (CMR) images is essential to diagnose and manage cardiovascular diseases. This study systematically evaluates the performance of five U-Net variants in cardiac MRI segmentation using the Automated Cardiac Diagnosis Challenge (ACDC) dataset and a hybrid loss function combining Cross-Entropy and dice losses. Among the variants, the Feature Pyramid U-Net achieved the best performance, with Dice coefficients of 0.9388 (Left Ventricle), 0.8759 (Right Ventricle), and 0.8426 (Myocardium), showcasing its superior ability to capture multi-scale features and segment complex anatomical structures. The comprehensive and standardized evaluation conducted in this study provides valuable insights into the strengths and limitations of these architectures for cardiac segmentation.

Original language	English
Title of host publication	2025 5th International Conference on Advanced Research in Computing (ICARC)
Publisher	IEEE Xplore
Pages	1-6
Number of pages	6
ISBN (Electronic)	979-8-3315-3098-3
ISBN (Print)	979-8-3315-3099-0
DOIs	https://doi.org/10.1109/ICARC64760.2025.10963166
Publication status	Published online - 16 Apr 2025
Event	5th International Conference on Advanced Research in Computing - Online Duration: 19 Feb 2025 → 20 Feb 2025 Conference number: 2025

Conference

Conference	5th International Conference on Advanced Research in Computing
Abbreviated title	ICARC
Period	19/02/25 → 20/02/25

Bibliographical note

Publisher Copyright:
© 2025 IEEE.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Artificial Intelligence
Cardiac MRI
Deep learning
Segmentation
U-Net
Deep Learning

Access to Document

10.1109/ICARC64760.2025.10963166

Initial submissionAuthor Accepted Manuscript, 986 KB

Cite this

@inproceedings{429244cd1ea940948c1bbc6ba07366cf,

title = "Cardiac MRI Segmentation of Ventricular Structures and Myocardium Using U-Net Variants",

abstract = "Accurate segmentation of the ventricular structures and myocardium from Cardiac Magnetic Resonance (CMR) images is essential to diagnose and manage cardiovascular diseases. This study systematically evaluates the performance of five U-Net variants in cardiac MRI segmentation using the Automated Cardiac Diagnosis Challenge (ACDC) dataset and a hybrid loss function combining Cross-Entropy and dice losses. Among the variants, the Feature Pyramid U-Net achieved the best performance, with Dice coefficients of 0.9388 (Left Ventricle), 0.8759 (Right Ventricle), and 0.8426 (Myocardium), showcasing its superior ability to capture multi-scale features and segment complex anatomical structures. The comprehensive and standardized evaluation conducted in this study provides valuable insights into the strengths and limitations of these architectures for cardiac segmentation.",

keywords = "Artificial Intelligence, Cardiac MRI, Deep learning, Segmentation, U-Net, Deep Learning",

author = "Chinthalanka Wijesinghe and Dulani Meedeniya and Pratheepan Yogarajah",

note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 5th International Conference on Advanced Research in Computing, ICARC ; Conference date: 19-02-2025 Through 20-02-2025",

year = "2025",

month = apr,

day = "16",

doi = "10.1109/ICARC64760.2025.10963166",

language = "English",

isbn = "979-8-3315-3099-0",

pages = "1--6",

booktitle = "2025 5th International Conference on Advanced Research in Computing (ICARC)",

publisher = "IEEE Xplore",

address = "United States",

}

TY - GEN

T1 - Cardiac MRI Segmentation of Ventricular Structures and Myocardium Using U-Net Variants

AU - Wijesinghe, Chinthalanka

AU - Meedeniya, Dulani

AU - Yogarajah, Pratheepan

N1 - Conference code: 2025

PY - 2025/4/16

Y1 - 2025/4/16

N2 - Accurate segmentation of the ventricular structures and myocardium from Cardiac Magnetic Resonance (CMR) images is essential to diagnose and manage cardiovascular diseases. This study systematically evaluates the performance of five U-Net variants in cardiac MRI segmentation using the Automated Cardiac Diagnosis Challenge (ACDC) dataset and a hybrid loss function combining Cross-Entropy and dice losses. Among the variants, the Feature Pyramid U-Net achieved the best performance, with Dice coefficients of 0.9388 (Left Ventricle), 0.8759 (Right Ventricle), and 0.8426 (Myocardium), showcasing its superior ability to capture multi-scale features and segment complex anatomical structures. The comprehensive and standardized evaluation conducted in this study provides valuable insights into the strengths and limitations of these architectures for cardiac segmentation.

AB - Accurate segmentation of the ventricular structures and myocardium from Cardiac Magnetic Resonance (CMR) images is essential to diagnose and manage cardiovascular diseases. This study systematically evaluates the performance of five U-Net variants in cardiac MRI segmentation using the Automated Cardiac Diagnosis Challenge (ACDC) dataset and a hybrid loss function combining Cross-Entropy and dice losses. Among the variants, the Feature Pyramid U-Net achieved the best performance, with Dice coefficients of 0.9388 (Left Ventricle), 0.8759 (Right Ventricle), and 0.8426 (Myocardium), showcasing its superior ability to capture multi-scale features and segment complex anatomical structures. The comprehensive and standardized evaluation conducted in this study provides valuable insights into the strengths and limitations of these architectures for cardiac segmentation.

KW - Artificial Intelligence

KW - Cardiac MRI

KW - Deep learning

KW - Segmentation

KW - U-Net

KW - Deep Learning

UR - https://www.scopus.com/pages/publications/105004561177

U2 - 10.1109/ICARC64760.2025.10963166

DO - 10.1109/ICARC64760.2025.10963166

M3 - Conference contribution

SN - 979-8-3315-3099-0

SP - 1

EP - 6

BT - 2025 5th International Conference on Advanced Research in Computing (ICARC)

PB - IEEE Xplore

T2 - 5th International Conference on Advanced Research in Computing

Y2 - 19 February 2025 through 20 February 2025

ER -

Cardiac MRI Segmentation of Ventricular Structures and Myocardium Using U-Net Variants

Abstract

Conference

Bibliographical note

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this