Home > Research > Publications & Outputs > UDT

Links

Text available via DOI:

View graph of relations

UDT: U-shaped deformable transformer for subarachnoid haemorrhage image segmentation

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published

Standard

UDT: U-shaped deformable transformer for subarachnoid haemorrhage image segmentation. / Xie, Wei; Jin, Lianghao; Hua, Shiqi et al.
In: CAAI Transactions on Intelligence Technology, Vol. 9, No. 3, 30.06.2024, p. 756-768.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Harvard

Xie, W, Jin, L, Hua, S, Sun, H, Sun, B, Tu, Z & Liu, J 2024, 'UDT: U-shaped deformable transformer for subarachnoid haemorrhage image segmentation', CAAI Transactions on Intelligence Technology, vol. 9, no. 3, pp. 756-768. https://doi.org/10.1049/cit2.12302

APA

Xie, W., Jin, L., Hua, S., Sun, H., Sun, B., Tu, Z., & Liu, J. (2024). UDT: U-shaped deformable transformer for subarachnoid haemorrhage image segmentation. CAAI Transactions on Intelligence Technology, 9(3), 756-768. https://doi.org/10.1049/cit2.12302

Vancouver

Xie W, Jin L, Hua S, Sun H, Sun B, Tu Z et al. UDT: U-shaped deformable transformer for subarachnoid haemorrhage image segmentation. CAAI Transactions on Intelligence Technology. 2024 Jun 30;9(3):756-768. Epub 2024 Mar 25. doi: 10.1049/cit2.12302

Author

Xie, Wei ; Jin, Lianghao ; Hua, Shiqi et al. / UDT : U-shaped deformable transformer for subarachnoid haemorrhage image segmentation. In: CAAI Transactions on Intelligence Technology. 2024 ; Vol. 9, No. 3. pp. 756-768.

Bibtex

@article{930b5c617f134c368184ca14d1f97a66,
title = "UDT: U-shaped deformable transformer for subarachnoid haemorrhage image segmentation",
abstract = "Subarachnoid haemorrhage (SAH), mostly caused by the rupture of intracranial aneurysm, is a common disease with a high fatality rate. SAH lesions are generally diffusely distributed, showing a variety of scales with irregular edges. The complex characteristics of lesions make SAH segmentation a challenging task. To cope with these difficulties, a u-shaped deformable transformer (UDT) is proposed for SAH segmentation. Specifically, first, a multi-scale deformable attention (MSDA) module is exploited to model the diffuseness and scale-variant characteristics of SAH lesions, where the MSDA module can fuse features in different scales and adjust the attention field of each element dynamically to generate discriminative multi-scale features. Second, the cross deformable attention-based skip connection (CDASC) module is designed to model the irregular edge characteristic of SAH lesions, where the CDASC module can utilise the spatial details from encoder features to refine the spatial information of decoder features. Third, the MSDA and CDASC modules are embedded into the backbone Res-UNet to construct the proposed UDT. Extensive experiments are conducted on the self-built SAH-CT dataset and two public medical datasets (GlaS and MoNuSeg). Experimental results show that the presented UDT achieves the state-of-the-art performance.",
keywords = "image segmentation, medical image processing",
author = "Wei Xie and Lianghao Jin and Shiqi Hua and Hao Sun and Bo Sun and Zhigang Tu and Jun Liu",
note = "Publisher Copyright: {\textcopyright} 2024 The Authors. CAAI Transactions on Intelligence Technology published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology and Chongqing University of Technology.",
year = "2024",
month = jun,
day = "30",
doi = "10.1049/cit2.12302",
language = "English",
volume = "9",
pages = "756--768",
journal = "CAAI Transactions on Intelligence Technology",
issn = "2468-6557",
publisher = "John Wiley & Sons Inc.",
number = "3",

}

RIS

TY - JOUR

T1 - UDT

T2 - U-shaped deformable transformer for subarachnoid haemorrhage image segmentation

AU - Xie, Wei

AU - Jin, Lianghao

AU - Hua, Shiqi

AU - Sun, Hao

AU - Sun, Bo

AU - Tu, Zhigang

AU - Liu, Jun

N1 - Publisher Copyright: © 2024 The Authors. CAAI Transactions on Intelligence Technology published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology and Chongqing University of Technology.

PY - 2024/6/30

Y1 - 2024/6/30

N2 - Subarachnoid haemorrhage (SAH), mostly caused by the rupture of intracranial aneurysm, is a common disease with a high fatality rate. SAH lesions are generally diffusely distributed, showing a variety of scales with irregular edges. The complex characteristics of lesions make SAH segmentation a challenging task. To cope with these difficulties, a u-shaped deformable transformer (UDT) is proposed for SAH segmentation. Specifically, first, a multi-scale deformable attention (MSDA) module is exploited to model the diffuseness and scale-variant characteristics of SAH lesions, where the MSDA module can fuse features in different scales and adjust the attention field of each element dynamically to generate discriminative multi-scale features. Second, the cross deformable attention-based skip connection (CDASC) module is designed to model the irregular edge characteristic of SAH lesions, where the CDASC module can utilise the spatial details from encoder features to refine the spatial information of decoder features. Third, the MSDA and CDASC modules are embedded into the backbone Res-UNet to construct the proposed UDT. Extensive experiments are conducted on the self-built SAH-CT dataset and two public medical datasets (GlaS and MoNuSeg). Experimental results show that the presented UDT achieves the state-of-the-art performance.

AB - Subarachnoid haemorrhage (SAH), mostly caused by the rupture of intracranial aneurysm, is a common disease with a high fatality rate. SAH lesions are generally diffusely distributed, showing a variety of scales with irregular edges. The complex characteristics of lesions make SAH segmentation a challenging task. To cope with these difficulties, a u-shaped deformable transformer (UDT) is proposed for SAH segmentation. Specifically, first, a multi-scale deformable attention (MSDA) module is exploited to model the diffuseness and scale-variant characteristics of SAH lesions, where the MSDA module can fuse features in different scales and adjust the attention field of each element dynamically to generate discriminative multi-scale features. Second, the cross deformable attention-based skip connection (CDASC) module is designed to model the irregular edge characteristic of SAH lesions, where the CDASC module can utilise the spatial details from encoder features to refine the spatial information of decoder features. Third, the MSDA and CDASC modules are embedded into the backbone Res-UNet to construct the proposed UDT. Extensive experiments are conducted on the self-built SAH-CT dataset and two public medical datasets (GlaS and MoNuSeg). Experimental results show that the presented UDT achieves the state-of-the-art performance.

KW - image segmentation

KW - medical image processing

U2 - 10.1049/cit2.12302

DO - 10.1049/cit2.12302

M3 - Journal article

AN - SCOPUS:85189502648

VL - 9

SP - 756

EP - 768

JO - CAAI Transactions on Intelligence Technology

JF - CAAI Transactions on Intelligence Technology

SN - 2468-6557

IS - 3

ER -