Cargando…

Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging

T(2)-weighted magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI) are essential components of cervical cancer diagnosis. However, combining these channels for the training of deep learning models is challenging due to image misalignment. Here, we propose a novel multi-head framewor...

Descripción completa

Detalles Bibliográficos
Autores principales: Kalantar, Reza, Curcean, Sebastian, Winfield, Jessica M., Lin, Gigin, Messiou, Christina, Blackledge, Matthew D., Koh, Dow-Mu
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2023
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10647438/
https://www.ncbi.nlm.nih.gov/pubmed/37958277
http://dx.doi.org/10.3390/diagnostics13213381
_version_ 1785135107206545408
author Kalantar, Reza
Curcean, Sebastian
Winfield, Jessica M.
Lin, Gigin
Messiou, Christina
Blackledge, Matthew D.
Koh, Dow-Mu
author_facet Kalantar, Reza
Curcean, Sebastian
Winfield, Jessica M.
Lin, Gigin
Messiou, Christina
Blackledge, Matthew D.
Koh, Dow-Mu
author_sort Kalantar, Reza
collection PubMed
description T(2)-weighted magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI) are essential components of cervical cancer diagnosis. However, combining these channels for the training of deep learning models is challenging due to image misalignment. Here, we propose a novel multi-head framework that uses dilated convolutions and shared residual connections for the separate encoding of multiparametric MRI images. We employ a residual U-Net model as a baseline, and perform a series of architectural experiments to evaluate the tumor segmentation performance based on multiparametric input channels and different feature encoding configurations. All experiments were performed on a cohort of 207 patients with locally advanced cervical cancer. Our proposed multi-head model using separate dilated encoding for T(2)W MRI and combined b1000 DWI and apparent diffusion coefficient (ADC) maps achieved the best median Dice similarity coefficient (DSC) score, 0.823 (confidence interval (CI), 0.595–0.797), outperforming the conventional multi-channel model, DSC 0.788 (95% CI, 0.568–0.776), although the difference was not statistically significant (p > 0.05). We investigated channel sensitivity using 3D GRAD-CAM and channel dropout, and highlighted the critical importance of T(2)W and ADC channels for accurate tumor segmentation. However, our results showed that b1000 DWI had a minor impact on the overall segmentation performance. We demonstrated that the use of separate dilated feature extractors and independent contextual learning improved the model’s ability to reduce the boundary effects and distortion of DWI, leading to improved segmentation performance. Our findings could have significant implications for the development of robust and generalizable models that can extend to other multi-modal segmentation applications.
format Online
Article
Text
id pubmed-10647438
institution National Center for Biotechnology Information
language English
publishDate 2023
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-106474382023-11-03 Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging Kalantar, Reza Curcean, Sebastian Winfield, Jessica M. Lin, Gigin Messiou, Christina Blackledge, Matthew D. Koh, Dow-Mu Diagnostics (Basel) Article T(2)-weighted magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI) are essential components of cervical cancer diagnosis. However, combining these channels for the training of deep learning models is challenging due to image misalignment. Here, we propose a novel multi-head framework that uses dilated convolutions and shared residual connections for the separate encoding of multiparametric MRI images. We employ a residual U-Net model as a baseline, and perform a series of architectural experiments to evaluate the tumor segmentation performance based on multiparametric input channels and different feature encoding configurations. All experiments were performed on a cohort of 207 patients with locally advanced cervical cancer. Our proposed multi-head model using separate dilated encoding for T(2)W MRI and combined b1000 DWI and apparent diffusion coefficient (ADC) maps achieved the best median Dice similarity coefficient (DSC) score, 0.823 (confidence interval (CI), 0.595–0.797), outperforming the conventional multi-channel model, DSC 0.788 (95% CI, 0.568–0.776), although the difference was not statistically significant (p > 0.05). We investigated channel sensitivity using 3D GRAD-CAM and channel dropout, and highlighted the critical importance of T(2)W and ADC channels for accurate tumor segmentation. However, our results showed that b1000 DWI had a minor impact on the overall segmentation performance. We demonstrated that the use of separate dilated feature extractors and independent contextual learning improved the model’s ability to reduce the boundary effects and distortion of DWI, leading to improved segmentation performance. Our findings could have significant implications for the development of robust and generalizable models that can extend to other multi-modal segmentation applications. MDPI 2023-11-03 /pmc/articles/PMC10647438/ /pubmed/37958277 http://dx.doi.org/10.3390/diagnostics13213381 Text en © 2023 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Kalantar, Reza
Curcean, Sebastian
Winfield, Jessica M.
Lin, Gigin
Messiou, Christina
Blackledge, Matthew D.
Koh, Dow-Mu
Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging
title Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging
title_full Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging
title_fullStr Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging
title_full_unstemmed Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging
title_short Deep Learning Framework with Multi-Head Dilated Encoders for Enhanced Segmentation of Cervical Cancer on Multiparametric Magnetic Resonance Imaging
title_sort deep learning framework with multi-head dilated encoders for enhanced segmentation of cervical cancer on multiparametric magnetic resonance imaging
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10647438/
https://www.ncbi.nlm.nih.gov/pubmed/37958277
http://dx.doi.org/10.3390/diagnostics13213381
work_keys_str_mv AT kalantarreza deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging
AT curceansebastian deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging
AT winfieldjessicam deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging
AT lingigin deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging
AT messiouchristina deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging
AT blackledgematthewd deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging
AT kohdowmu deeplearningframeworkwithmultiheaddilatedencodersforenhancedsegmentationofcervicalcanceronmultiparametricmagneticresonanceimaging