BM-Seg: A new bone metastases segmentation ...
Type de document :
Article dans une revue scientifique: Article original
Titre :
BM-Seg: A new bone metastases segmentation dataset and ensemble of CNN-based segmentation approach
Auteur(s) :
Afnouch, Marwa [Auteur correspondant]
COMmunications NUMériques - IEMN [COMNUM - IEMN]
Institut d’Électronique, de Microélectronique et de Nanotechnologie - UMR 8520 [IEMN]
Gaddour, Olfa [Auteur correspondant]
Hentati, Yosr [Auteur correspondant]
Hedi Chaker Hospital [Sfax] [CHU Sfax]
Bougourzi, Fares [Auteur correspondant]
Istituto di Scienze Applicate e Sistemi Intelligenti “Eduardo Caianiello” [ISASI]
Abid, Mohamed [Auteur correspondant]
Hedi Chaker Hospital [Sfax] [CHU Sfax]
Alouani, Lihsen [Auteur correspondant]
Queen's University [Belfast] [QUB]
COMmunications NUMériques - IEMN [COMNUM - IEMN]
Institut d’Électronique, de Microélectronique et de Nanotechnologie - UMR 8520 [IEMN]
Tahleb Ahmed, Abdelmalik [Auteur correspondant]
Institut d’Électronique, de Microélectronique et de Nanotechnologie - UMR 8520 [IEMN]
COMmunications NUMériques - IEMN [COMNUM - IEMN]
COMmunications NUMériques - IEMN [COMNUM - IEMN]
Institut d’Électronique, de Microélectronique et de Nanotechnologie - UMR 8520 [IEMN]
Gaddour, Olfa [Auteur correspondant]
Hentati, Yosr [Auteur correspondant]
Hedi Chaker Hospital [Sfax] [CHU Sfax]
Bougourzi, Fares [Auteur correspondant]
Istituto di Scienze Applicate e Sistemi Intelligenti “Eduardo Caianiello” [ISASI]
Abid, Mohamed [Auteur correspondant]
Hedi Chaker Hospital [Sfax] [CHU Sfax]
Alouani, Lihsen [Auteur correspondant]
Queen's University [Belfast] [QUB]
COMmunications NUMériques - IEMN [COMNUM - IEMN]
Institut d’Électronique, de Microélectronique et de Nanotechnologie - UMR 8520 [IEMN]
Tahleb Ahmed, Abdelmalik [Auteur correspondant]
Institut d’Électronique, de Microélectronique et de Nanotechnologie - UMR 8520 [IEMN]
COMmunications NUMériques - IEMN [COMNUM - IEMN]
Titre de la revue :
Expert systems with applications
Pagination :
120376
Éditeur :
Elsevier
Date de publication :
2023-10
ISSN :
0957-4174
Mot(s)-clé(s) en anglais :
BM-Seg
Bone metastases
CT-scans
Semantic segmentation
Generation
Convolutional neural network
Bone metastases
CT-scans
Semantic segmentation
Generation
Convolutional neural network
Discipline(s) HAL :
Physique [physics]
Sciences de l'ingénieur [physics]
Sciences de l'ingénieur [physics]
Résumé en anglais : [en]
A B S T R A C TIn recent years, Machine Learning approaches (ML) have shown promising results in addressing many tasks inmedical image analysis. In particular, the analysis of Bone Metastases (BM) has attracted considerable ...
Lire la suite >A B S T R A C TIn recent years, Machine Learning approaches (ML) have shown promising results in addressing many tasks inmedical image analysis. In particular, the analysis of Bone Metastases (BM) has attracted considerable interestfrom both the medical and computer vision communities due to its critical and challenging aspect. Despitethe research efforts, the detection of BM is still an open problem, mainly due to the lack of available datasets.This is due to two main obstacles: (i) the enormous time required for data collection and annotation, and(ii) privacy constraints. To overcome these challenges, we propose BM-Seg, a new dataset for segmenting BMfrom CT-scans. Our BM-Seg dataset consists of 1517 CT images from 23 patients where BM and bone regionswere labeled by three radiologists. BM-Seg is constructed to cover the diversity of bone metastases in termsof location, organ and severity.We also propose a new CNN-based approach to segmentation of BM, presenting two main contributions.First, we introduce Hybrid-AttUnet++, a new Unet++ derived architecture with dual decoders that performssegmentation of BM and bone regions simultaneously. Second, we use an ensemble of trained HybridAttUnet++ models (EH-AttUnet++) to optimize segmentation performance. Our experiments show that theEH-AttUnet++ architecture achieves better performance compared to state-of-the-art approaches for variousevaluation metrics. The purpose of this work is to provide a benchmark dataset with new state-of-the-artperformance in bone metastasis segmentation. This will facilitate further research in this area and help to putautomatic detection and segmentation of bone metastases into practice.Lire moins >
Lire la suite >A B S T R A C TIn recent years, Machine Learning approaches (ML) have shown promising results in addressing many tasks inmedical image analysis. In particular, the analysis of Bone Metastases (BM) has attracted considerable interestfrom both the medical and computer vision communities due to its critical and challenging aspect. Despitethe research efforts, the detection of BM is still an open problem, mainly due to the lack of available datasets.This is due to two main obstacles: (i) the enormous time required for data collection and annotation, and(ii) privacy constraints. To overcome these challenges, we propose BM-Seg, a new dataset for segmenting BMfrom CT-scans. Our BM-Seg dataset consists of 1517 CT images from 23 patients where BM and bone regionswere labeled by three radiologists. BM-Seg is constructed to cover the diversity of bone metastases in termsof location, organ and severity.We also propose a new CNN-based approach to segmentation of BM, presenting two main contributions.First, we introduce Hybrid-AttUnet++, a new Unet++ derived architecture with dual decoders that performssegmentation of BM and bone regions simultaneously. Second, we use an ensemble of trained HybridAttUnet++ models (EH-AttUnet++) to optimize segmentation performance. Our experiments show that theEH-AttUnet++ architecture achieves better performance compared to state-of-the-art approaches for variousevaluation metrics. The purpose of this work is to provide a benchmark dataset with new state-of-the-artperformance in bone metastasis segmentation. This will facilitate further research in this area and help to putautomatic detection and segmentation of bone metastases into practice.Lire moins >
Langue :
Anglais
Comité de lecture :
Oui
Audience :
Internationale
Vulgarisation :
Non
Source :