A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy

Abstract This study developed and evaluated an automatic segmentation model based on the Mamba framework (AM-UNet) for rapid and precise delineation of high-risk clinical target volume (HRCTV) and organs at risk (OARs) in cervical cancer brachytherapy. Using 694 CT scans from 179 cervical cancer pat...

Full description

Saved in:
Bibliographic Details
Main Authors: Lele Zang, Jing Liu, Huiqi Zhang, Shitao Zhu, Mingxuan Zhu, Yuqin Wang, Yaxin Kang, Jihong Chen, Qin Xu
Format: Article
Language:English
Published: Nature Portfolio 2025-03-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-94431-1
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850064023883612160
author Lele Zang
Jing Liu
Huiqi Zhang
Shitao Zhu
Mingxuan Zhu
Yuqin Wang
Yaxin Kang
Jihong Chen
Qin Xu
author_facet Lele Zang
Jing Liu
Huiqi Zhang
Shitao Zhu
Mingxuan Zhu
Yuqin Wang
Yaxin Kang
Jihong Chen
Qin Xu
author_sort Lele Zang
collection DOAJ
description Abstract This study developed and evaluated an automatic segmentation model based on the Mamba framework (AM-UNet) for rapid and precise delineation of high-risk clinical target volume (HRCTV) and organs at risk (OARs) in cervical cancer brachytherapy. Using 694 CT scans from 179 cervical cancer patients, the performance of five models (AM-UNet, UNet, DeepLab V3, UNETR and nnU-Net) was compared. The models were assessed using the Dice similarity coefficient (DSC), 95% Hausdorff distance (HD95), and dose-volume index (DVI). AM-UNet achieved mean DSCs of 0.862, 0.937, 0.823, and 0.725 for HRCTV, bladder, rectum, and sigmoid, respectively. Subjective evaluations showed 93.07% of AM-UNet predicted HRCTV were rated as clinically acceptable or needing minor adjustments, with no unacceptable cases. Dosimetric differences between AM-UNet-generated and manually delineated contours were within 1%, highlighting its potential for improving clinical workflows in brachytherapy.
format Article
id doaj-art-7a966a734e9940d68ef4f0a6dbaa84c0
institution DOAJ
issn 2045-2322
language English
publishDate 2025-03-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-7a966a734e9940d68ef4f0a6dbaa84c02025-08-20T02:49:26ZengNature PortfolioScientific Reports2045-23222025-03-0115111010.1038/s41598-025-94431-1A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapyLele Zang0Jing Liu1Huiqi Zhang2Shitao Zhu3Mingxuan Zhu4Yuqin Wang5Yaxin Kang6Jihong Chen7Qin Xu8Department of Gynecology, Clinical Oncology School of Fujian Medical University, Fujian Cancer HospitalDepartment of Gynecology, Clinical Oncology School of Fujian Medical University, Fujian Cancer HospitalDepartment of Radiation Oncology, Fujian Cancer Hospital, Clinical Oncology School of Fujian Medical UniversityCollege of Computer and Data Science, Fuzhou UniversityDepartment of Radiation Oncology, Fujian Cancer Hospital, Clinical Oncology School of Fujian Medical UniversityDepartment of Radiation Oncology, Fujian Cancer Hospital, Clinical Oncology School of Fujian Medical UniversityDepartment of Radiation Oncology, Fujian Cancer Hospital, Clinical Oncology School of Fujian Medical UniversityDepartment of Radiation Oncology, Clinical Oncology School of Fujian Medical University, Fujian Cancer HospitalDepartment of Gynecology, Clinical Oncology School of Fujian Medical University, Fujian Cancer HospitalAbstract This study developed and evaluated an automatic segmentation model based on the Mamba framework (AM-UNet) for rapid and precise delineation of high-risk clinical target volume (HRCTV) and organs at risk (OARs) in cervical cancer brachytherapy. Using 694 CT scans from 179 cervical cancer patients, the performance of five models (AM-UNet, UNet, DeepLab V3, UNETR and nnU-Net) was compared. The models were assessed using the Dice similarity coefficient (DSC), 95% Hausdorff distance (HD95), and dose-volume index (DVI). AM-UNet achieved mean DSCs of 0.862, 0.937, 0.823, and 0.725 for HRCTV, bladder, rectum, and sigmoid, respectively. Subjective evaluations showed 93.07% of AM-UNet predicted HRCTV were rated as clinically acceptable or needing minor adjustments, with no unacceptable cases. Dosimetric differences between AM-UNet-generated and manually delineated contours were within 1%, highlighting its potential for improving clinical workflows in brachytherapy.https://doi.org/10.1038/s41598-025-94431-1Cervical cancerBrachytherapyAuto-segmentationDeep learningComputed
spellingShingle Lele Zang
Jing Liu
Huiqi Zhang
Shitao Zhu
Mingxuan Zhu
Yuqin Wang
Yaxin Kang
Jihong Chen
Qin Xu
A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy
Scientific Reports
Cervical cancer
Brachytherapy
Auto-segmentation
Deep learning
Computed
title A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy
title_full A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy
title_fullStr A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy
title_full_unstemmed A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy
title_short A deep learning model based on Mamba for automatic segmentation in cervical cancer brachytherapy
title_sort deep learning model based on mamba for automatic segmentation in cervical cancer brachytherapy
topic Cervical cancer
Brachytherapy
Auto-segmentation
Deep learning
Computed
url https://doi.org/10.1038/s41598-025-94431-1
work_keys_str_mv AT lelezang adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT jingliu adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT huiqizhang adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT shitaozhu adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT mingxuanzhu adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT yuqinwang adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT yaxinkang adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT jihongchen adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT qinxu adeeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT lelezang deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT jingliu deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT huiqizhang deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT shitaozhu deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT mingxuanzhu deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT yuqinwang deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT yaxinkang deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT jihongchen deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy
AT qinxu deeplearningmodelbasedonmambaforautomaticsegmentationincervicalcancerbrachytherapy