An Evidential Mask Transformer for Left Atrium Segmentation

Fatmatülzehra Uslu

Article Details

An Evidential Mask Transformer for Left Atrium Segmentation

Authors:

Fatmatülzehra Uslu

Publisher:

Cukurova University

Main Topics:

Biyomedikal Mühendisliği (Diğer)

Journal:

Çukurova Üniversitesi Mühendislik Fakültesi Dergisi Vol.39 Issue 3, pp.639-646

Document Type:

Article

Publication Date:

3/2024

Document Language:

English

Document ID (Org No.):

2001329022

pISSN:

2757-9255

Keywords:

Abstract:

 The segmentation of the left atrium LA is required to calculate the clinical parameters of the LA, to identify diseases related to its remodeling. Generally, convolutional networks have been used for this task. However, their performance may be limited as a result of the use of local convolution operations for feature extraction. Also, such models usually need extra steps to provide uncertainty maps such as multiple forward passes for Monte Carlo dropouts or training multiple models for ensemble learning. To address these issues, we adapt mask transformers for LA segmentation which effectively use both local and global information, and train them with evidential learning to generate uncertainty maps from the learned Dirichlet distribution, with a single forward pass. We validated our approach on the STACOM 2013 dataset and found that our method can produce better segmentation performance than baseline models, and can identify locations our models responses are not trustable.

 The segmentation of the left atrium LA is required to calculate the clinical parameters of the LA, to identify diseases related to its remodeling. Generally, convolutional networks have been used for this task. However, their performance may be limited as a result of the use of local convolution operations for feature extraction. Also, such models usually need extra steps to provide uncertainty maps such as multiple forward passes for Monte Carlo dropouts or training multiple models for ensemble learning. To address these issues, we adapt mask transformers for LA segmentation which effectively use both local and global information, and train them with evidential learning to generate uncertainty maps from the learned Dirichlet distribution, with a single forward pass. We validated our approach on the STACOM 2013 dataset and found that our method can produce better segmentation performance than baseline models, and can identify locations our models responses are not trustable.

 The segmentation of the left atrium LA is required to calculate the clinical parameters of the LA, to identify diseases related to its remodeling. Generally, convolutional networks have been used for this task. However, their performance may be limited as a result of the use of local convolution operations for feature extraction. Also, such models usually need extra steps to provide uncertainty maps such as multiple forward passes for Monte Carlo dropouts or training multiple models for ensemble learning. To address these issues, we adapt mask transformers for LA segmentation which effectively use both local and global information, and train them with evidential learning to generate uncertainty maps from the learned Dirichlet distribution, with a single forward pass. We validated our approach on the STACOM 2013 dataset and found that our method can produce better segmentation performance than baseline models, and can identify locations our models responses are not trustable.

 Sol kulakçığın yeniden şekillenmesine sebep olan hastalıklarının tanısının konulabilmesi için, sol kulakçığın bölütlenmesi gerekmektedir. Bu amaçla, genel olarak, konvolusyenel ağlar kullanılmaktadır. Fakat bu modellerin performansı, yerel hesaplama yapmaları nedeniyle düşük olabilir. Belirsizlik haritaları üretebilmeleri için, Monte Karlo dropout ya da çoklu model eğitimi ensemble gibi yaklaşımlara ihtiyaç duyulur. Bu problemleri gidermek için, yerel ve global bilgiyi bir arada kullanan, maske dönüştürücü modelleri, sol kulakçık bölütlenmesi için adapte ettik. Belirsizlik haritalarını elde etmek için de bu modeller, kanıtsal öğrenme ile eğitildi. Böylece, öğrenilen Dirichlet dağılımı kullanılarak, tek adımda belirsizlik haritaları elde edilebildi. Öne sürülen yaklaşım, STACOM 2013 veri setinde test edildi ve karşılaştırılan modellerden daha başarılı performans gösterdiği gözlemlendi. Üretilen belirsizlik haritalarının, modelin kararsız olduğu yerlerde yüksek belirsizlik gösterdiği gözlemlendi.

Article Details

An Evidential Mask Transformer for Left Atrium Segmentation

Authors:

Fatmatülzehra Uslu

Publisher:

Cukurova University

Main Topics:

Biyomedikal Mühendisliği (Diğer)

Journal:

Çukurova Üniversitesi Mühendislik Fakültesi Dergisi Vol.39 Issue 3, pp.639-646

Document Type:

Article

Publication Date:

3/2024

Document Language:

English

Document ID (Org No.):

2001329022

pISSN:

2757-9255

Mask transformer

Image segmentation

Uncertainty

MRI images

Evidential learningMaske dönüştürücü

Görüntü bölütleme

Belirsizlik

Manyetik rezonans görüntüleri

Kanıtsal öğrenme

Issues

Article Details

Authors: Fatmatülzehra Uslu

Publisher: Cukurova University

Main Topics: Biyomedikal Mühendisliği (Diğer)

Journal: Çukurova Üniversitesi Mühendislik Fakültesi Dergisi Vol.39 Issue 3, pp.639-646

Document Type: Article

Publication Date: 3/2024

Document Language: English

Document ID (Org No.): 2001329022

pISSN: 2757-9255

Issues

Authors:

Fatmatülzehra Uslu

Publisher:

Cukurova University

Main Topics:

Biyomedikal Mühendisliği (Diğer)

Journal:

Çukurova Üniversitesi Mühendislik Fakültesi Dergisi Vol.39 Issue 3, pp.639-646

Document Type:

Article

Publication Date:

3/2024

Document Language:

English

Document ID (Org No.):

2001329022

pISSN:

2757-9255