Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation

Abstract: One of the most widely used treatments for cancer of the gastrointestinal (GI) tract is radiotherapy, which requires manual segmentation of the affected organs to deliver radiation without affecting healthy cells. Deep learning techniques have been used, especially variants of U-Net, to automate the organ segmentation process, increasing the efficiency of medical treatment. However, the effective segmentation of the GI tract organs remains an open research problem due to their high capacity to deform because of body movement and respiratory function. This work proposes a methodology that develops a weighted ensemble integrating U-Net++ models and Hidden Markov Models (2D-HMM) for semantic segmentation of the stomach and bowels. Our empirical evaluation reports a score of 0.811 for the Dice coefficient using Leave-One-Out Cross-Validation, which provides robustness to the results.

Saved in:
Bibliographic Details
Main Authors: Ramírez-Sánchez,Jairo Enrique, Martínez-Barrón,Pedro A., Medina-Aguilar,Hannia, Sánchez-Nigenda,Romeo
Format: Digital revista
Language:English
Published: Instituto Politécnico Nacional, Centro de Investigación en Computación 2023
Online Access:http://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S1405-55462023000400991
Tags: Add Tag
No Tags, Be the first to tag this record!
id oai:scielo:S1405-55462023000400991
record_format ojs
spelling oai:scielo:S1405-554620230004009912024-05-17Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image SegmentationRamírez-Sánchez,Jairo EnriqueMartínez-Barrón,Pedro A.Medina-Aguilar,HanniaSánchez-Nigenda,Romeo Image segmentation U-NET architecture machine learning hidden Markov models Abstract: One of the most widely used treatments for cancer of the gastrointestinal (GI) tract is radiotherapy, which requires manual segmentation of the affected organs to deliver radiation without affecting healthy cells. Deep learning techniques have been used, especially variants of U-Net, to automate the organ segmentation process, increasing the efficiency of medical treatment. However, the effective segmentation of the GI tract organs remains an open research problem due to their high capacity to deform because of body movement and respiratory function. This work proposes a methodology that develops a weighted ensemble integrating U-Net++ models and Hidden Markov Models (2D-HMM) for semantic segmentation of the stomach and bowels. Our empirical evaluation reports a score of 0.811 for the Dice coefficient using Leave-One-Out Cross-Validation, which provides robustness to the results.info:eu-repo/semantics/openAccessInstituto Politécnico Nacional, Centro de Investigación en ComputaciónComputación y Sistemas v.27 n.4 20232023-12-01info:eu-repo/semantics/articletext/htmlhttp://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S1405-55462023000400991en10.13053/cys-27-4-4771
institution SCIELO
collection OJS
country México
countrycode MX
component Revista
access En linea
databasecode rev-scielo-mx
tag revista
region America del Norte
libraryname SciELO
language English
format Digital
author Ramírez-Sánchez,Jairo Enrique
Martínez-Barrón,Pedro A.
Medina-Aguilar,Hannia
Sánchez-Nigenda,Romeo
spellingShingle Ramírez-Sánchez,Jairo Enrique
Martínez-Barrón,Pedro A.
Medina-Aguilar,Hannia
Sánchez-Nigenda,Romeo
Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation
author_facet Ramírez-Sánchez,Jairo Enrique
Martínez-Barrón,Pedro A.
Medina-Aguilar,Hannia
Sánchez-Nigenda,Romeo
author_sort Ramírez-Sánchez,Jairo Enrique
title Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation
title_short Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation
title_full Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation
title_fullStr Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation
title_full_unstemmed Weighted U-NET++ and 2D-HMM Ensemble for Gastrointestinal Image Segmentation
title_sort weighted u-net++ and 2d-hmm ensemble for gastrointestinal image segmentation
description Abstract: One of the most widely used treatments for cancer of the gastrointestinal (GI) tract is radiotherapy, which requires manual segmentation of the affected organs to deliver radiation without affecting healthy cells. Deep learning techniques have been used, especially variants of U-Net, to automate the organ segmentation process, increasing the efficiency of medical treatment. However, the effective segmentation of the GI tract organs remains an open research problem due to their high capacity to deform because of body movement and respiratory function. This work proposes a methodology that develops a weighted ensemble integrating U-Net++ models and Hidden Markov Models (2D-HMM) for semantic segmentation of the stomach and bowels. Our empirical evaluation reports a score of 0.811 for the Dice coefficient using Leave-One-Out Cross-Validation, which provides robustness to the results.
publisher Instituto Politécnico Nacional, Centro de Investigación en Computación
publishDate 2023
url http://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S1405-55462023000400991
work_keys_str_mv AT ramirezsanchezjairoenrique weightedunetand2dhmmensembleforgastrointestinalimagesegmentation
AT martinezbarronpedroa weightedunetand2dhmmensembleforgastrointestinalimagesegmentation
AT medinaaguilarhannia weightedunetand2dhmmensembleforgastrointestinalimagesegmentation
AT sancheznigendaromeo weightedunetand2dhmmensembleforgastrointestinalimagesegmentation
_version_ 1802825334839574528