MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging

Abstract Background Single-photon emission computed tomography (SPECT) plays a crucial role in detecting bone metastases from lung cancer. However, its low spatial resolution and lesion similarity to benign structures present significant challenges for accurate segmentation, especially for lesions o...

Full description

Saved in:
Bibliographic Details
Main Authors: Yusheng Wu, Qiang Lin, Yang He, XianWu Zeng, Yongchun Cao, ZhengXing Man, Caihong Liu, Yusheng Hao, Zhengqi Cai, Jinshui Ji, Xiaodi Huang
Format: Article
Language:English
Published: SpringerOpen 2025-07-01
Series:EJNMMI Physics
Subjects:
Online Access:https://doi.org/10.1186/s40658-025-00785-w
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849761178510688256
author Yusheng Wu
Qiang Lin
Yang He
XianWu Zeng
Yongchun Cao
ZhengXing Man
Caihong Liu
Yusheng Hao
Zhengqi Cai
Jinshui Ji
Xiaodi Huang
author_facet Yusheng Wu
Qiang Lin
Yang He
XianWu Zeng
Yongchun Cao
ZhengXing Man
Caihong Liu
Yusheng Hao
Zhengqi Cai
Jinshui Ji
Xiaodi Huang
author_sort Yusheng Wu
collection DOAJ
description Abstract Background Single-photon emission computed tomography (SPECT) plays a crucial role in detecting bone metastases from lung cancer. However, its low spatial resolution and lesion similarity to benign structures present significant challenges for accurate segmentation, especially for lesions of varying sizes. Methods We propose a deep learning-based segmentation framework that integrates conditional adversarial learning with a multi-scale feature extraction generator. The generator employs cascade dilated convolutions, multi-scale modules, and deep supervision, while the discriminator utilizes multi-scale L1 loss computed on image-mask pairs to guide segmentation learning. Results The proposed model was evaluated on a dataset of 286 clinically annotated SPECT scintigrams. It achieved a Dice Similarity Coefficient (DSC) of 0.6671, precision of 0.7228, and recall of 0.6196 — outperforming both classical and recent adversarial segmentation models in multi-scale lesion detection, especially for small and clustered lesions. Conclusion Our results demonstrate that the integration of multi-scale feature learning with adversarial supervision significantly improves the segmentation of bone metastasis in SPECT imaging. This approach shows potential for clinical decision support in the management of lung cancer.
format Article
id doaj-art-8035f63cf06847c8825ffaa18d52c49e
institution DOAJ
issn 2197-7364
language English
publishDate 2025-07-01
publisher SpringerOpen
record_format Article
series EJNMMI Physics
spelling doaj-art-8035f63cf06847c8825ffaa18d52c49e2025-08-20T03:06:06ZengSpringerOpenEJNMMI Physics2197-73642025-07-0112112110.1186/s40658-025-00785-wMSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imagingYusheng Wu0Qiang Lin1Yang He2XianWu Zeng3Yongchun Cao4ZhengXing Man5Caihong Liu6Yusheng Hao7Zhengqi Cai8Jinshui Ji9Xiaodi Huang10Key Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversityKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversitySchool of Mathematics and Computer Science, Northwest Minzu UniversityDepartment of Nuclear Medicine, Gansu Provincial Cancer HospitalKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversityKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversityKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversityKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversityKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversityKey Laboratory of China’s Ethnic Languages and Information Technology of Ministry of Education, Northwest Minzu UniversitySchool of Computing, Mathematics and Engineering, Charles Sturt UniversityAbstract Background Single-photon emission computed tomography (SPECT) plays a crucial role in detecting bone metastases from lung cancer. However, its low spatial resolution and lesion similarity to benign structures present significant challenges for accurate segmentation, especially for lesions of varying sizes. Methods We propose a deep learning-based segmentation framework that integrates conditional adversarial learning with a multi-scale feature extraction generator. The generator employs cascade dilated convolutions, multi-scale modules, and deep supervision, while the discriminator utilizes multi-scale L1 loss computed on image-mask pairs to guide segmentation learning. Results The proposed model was evaluated on a dataset of 286 clinically annotated SPECT scintigrams. It achieved a Dice Similarity Coefficient (DSC) of 0.6671, precision of 0.7228, and recall of 0.6196 — outperforming both classical and recent adversarial segmentation models in multi-scale lesion detection, especially for small and clustered lesions. Conclusion Our results demonstrate that the integration of multi-scale feature learning with adversarial supervision significantly improves the segmentation of bone metastasis in SPECT imaging. This approach shows potential for clinical decision support in the management of lung cancer.https://doi.org/10.1186/s40658-025-00785-wSPECT imageBone metastasisMultiscale segmentationAdversarial learning
spellingShingle Yusheng Wu
Qiang Lin
Yang He
XianWu Zeng
Yongchun Cao
ZhengXing Man
Caihong Liu
Yusheng Hao
Zhengqi Cai
Jinshui Ji
Xiaodi Huang
MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging
EJNMMI Physics
SPECT image
Bone metastasis
Multiscale segmentation
Adversarial learning
title MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging
title_full MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging
title_fullStr MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging
title_full_unstemmed MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging
title_short MSA-Net: a multi-scale and adversarial learning network for segmenting bone metastases in low-resolution SPECT imaging
title_sort msa net a multi scale and adversarial learning network for segmenting bone metastases in low resolution spect imaging
topic SPECT image
Bone metastasis
Multiscale segmentation
Adversarial learning
url https://doi.org/10.1186/s40658-025-00785-w
work_keys_str_mv AT yushengwu msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT qianglin msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT yanghe msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT xianwuzeng msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT yongchuncao msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT zhengxingman msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT caihongliu msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT yushenghao msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT zhengqicai msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT jinshuiji msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging
AT xiaodihuang msanetamultiscaleandadversariallearningnetworkforsegmentingbonemetastasesinlowresolutionspectimaging