SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation

Radiological diagnosis of lung cavities (LCs) is the key to identifying tuberculosis (TB). Conventional deep learning methods rely on a large amount of accurate pixel-level data to segment LCs. This process is timeconsuming and laborious, especially for those subtle LCs. To address such challenges,...

Full description

Bibliographic Details
Main Authors: Zhuoyi, Tan, Hizmawati, Madzin, Bahari, Norafida, Rahmita, Wirza OK Rahmat, Fatimah, Khalid, Puteri, Suhaiza Sulaiman
Format: Article
Language:English
Published: Elsevier 2024
Online Access:http://psasir.upm.edu.my/id/eprint/111381/
http://psasir.upm.edu.my/id/eprint/111381/1/SwinUNeLCsT.pdf
_version_ 1848865671816937472
author Zhuoyi, Tan
Hizmawati, Madzin
Bahari, Norafida
Rahmita, Wirza OK Rahmat
Fatimah, Khalid
Puteri, Suhaiza Sulaiman
author_facet Zhuoyi, Tan
Hizmawati, Madzin
Bahari, Norafida
Rahmita, Wirza OK Rahmat
Fatimah, Khalid
Puteri, Suhaiza Sulaiman
author_sort Zhuoyi, Tan
building UPM Institutional Repository
collection Online Access
description Radiological diagnosis of lung cavities (LCs) is the key to identifying tuberculosis (TB). Conventional deep learning methods rely on a large amount of accurate pixel-level data to segment LCs. This process is timeconsuming and laborious, especially for those subtle LCs. To address such challenges, firstly, we introduce a novel 3D TB LCs imaging convolutional neural network (CNN)-transformer hybrid model (SwinUNeLCsT). The core idea of SwinUNeLCsT is to combine local details and global dependencies for TB CT scan image feature representation to effectively improve the recognition ability of LCs. Secondly, to reduce the dependence on accurate pixel-level annotations, we design an end-to-end LCs weakly supervised semantic segmentation (WSSS) framework. Through this framework, radiologists need only to classify the number and the approximate location (e.g., left lung, right lung, or both) of LCs in the CT scan to achieve efficient segmentation of the LCs. This process eliminates the need for meticulously drawing boundaries, greatly reducing the cost of annotation. Extensive experimental results show that SwinUNeLCsT outperforms currently popular medical 3D segmentation methods in the supervised semantic segmentation paradigm. Meanwhile, our WSSS framework based on SwinUNeLCsT also performs best among the existing state-of-the-art medical 3D WSSS methods.
first_indexed 2025-11-15T14:08:25Z
format Article
id upm-111381
institution Universiti Putra Malaysia
institution_category Local University
language English
last_indexed 2025-11-15T14:08:25Z
publishDate 2024
publisher Elsevier
recordtype eprints
repository_type Digital Repository
spelling upm-1113812024-06-27T16:14:34Z http://psasir.upm.edu.my/id/eprint/111381/ SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation Zhuoyi, Tan Hizmawati, Madzin Bahari, Norafida Rahmita, Wirza OK Rahmat Fatimah, Khalid Puteri, Suhaiza Sulaiman Radiological diagnosis of lung cavities (LCs) is the key to identifying tuberculosis (TB). Conventional deep learning methods rely on a large amount of accurate pixel-level data to segment LCs. This process is timeconsuming and laborious, especially for those subtle LCs. To address such challenges, firstly, we introduce a novel 3D TB LCs imaging convolutional neural network (CNN)-transformer hybrid model (SwinUNeLCsT). The core idea of SwinUNeLCsT is to combine local details and global dependencies for TB CT scan image feature representation to effectively improve the recognition ability of LCs. Secondly, to reduce the dependence on accurate pixel-level annotations, we design an end-to-end LCs weakly supervised semantic segmentation (WSSS) framework. Through this framework, radiologists need only to classify the number and the approximate location (e.g., left lung, right lung, or both) of LCs in the CT scan to achieve efficient segmentation of the LCs. This process eliminates the need for meticulously drawing boundaries, greatly reducing the cost of annotation. Extensive experimental results show that SwinUNeLCsT outperforms currently popular medical 3D segmentation methods in the supervised semantic segmentation paradigm. Meanwhile, our WSSS framework based on SwinUNeLCsT also performs best among the existing state-of-the-art medical 3D WSSS methods. Elsevier 2024 Article PeerReviewed text en http://psasir.upm.edu.my/id/eprint/111381/1/SwinUNeLCsT.pdf Zhuoyi, Tan and Hizmawati, Madzin and Bahari, Norafida and Rahmita, Wirza OK Rahmat and Fatimah, Khalid and Puteri, Suhaiza Sulaiman (2024) SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation. Journal of King Saud University - Computer and Information Sciences, 36 (4). art. no. 102012. pp. 1-15. ISSN 1319-1578; ESSN: 2213-1248 https://www.sciencedirect.com/science/article/pii/S1319157824001010 10.1016/j.jksuci.2024.102012
spellingShingle Zhuoyi, Tan
Hizmawati, Madzin
Bahari, Norafida
Rahmita, Wirza OK Rahmat
Fatimah, Khalid
Puteri, Suhaiza Sulaiman
SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
title SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
title_full SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
title_fullStr SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
title_full_unstemmed SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
title_short SwinUNeLCsT: Global–local spatial representation learning with hybrid CNN–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
title_sort swinunelcst: global–local spatial representation learning with hybrid cnn–transformer for efficient tuberculosis lung cavity weakly supervised semantic segmentation
url http://psasir.upm.edu.my/id/eprint/111381/
http://psasir.upm.edu.my/id/eprint/111381/
http://psasir.upm.edu.my/id/eprint/111381/
http://psasir.upm.edu.my/id/eprint/111381/1/SwinUNeLCsT.pdf