A weakly-supervised follicle segmentation method in ultrasound images

Abstract Accurate follicle segmentation in ultrasound images is crucial for monitoring follicle development, a key factor in fertility treatments. However, obtaining pixel-level annotations for fully supervised instance segmentation is often impractical due to time and workload constraints. This pap...

Full description

Saved in:
Bibliographic Details
Main Authors: Guanyu Liu, Weihong Huang, Yanping Li, Qiong Zhang, Jing Fu, Hongying Tang, Jia Huang, Zhongteng Zhang, Lei Zhang, Yu Wang, Jianzhong Hu
Format: Article
Language:English
Published: Nature Portfolio 2025-04-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-95957-0
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849713173083455488
author Guanyu Liu
Weihong Huang
Yanping Li
Qiong Zhang
Jing Fu
Hongying Tang
Jia Huang
Zhongteng Zhang
Lei Zhang
Yu Wang
Jianzhong Hu
author_facet Guanyu Liu
Weihong Huang
Yanping Li
Qiong Zhang
Jing Fu
Hongying Tang
Jia Huang
Zhongteng Zhang
Lei Zhang
Yu Wang
Jianzhong Hu
author_sort Guanyu Liu
collection DOAJ
description Abstract Accurate follicle segmentation in ultrasound images is crucial for monitoring follicle development, a key factor in fertility treatments. However, obtaining pixel-level annotations for fully supervised instance segmentation is often impractical due to time and workload constraints. This paper presents a weakly supervised instance segmentation method that leverages bounding boxes as approximate annotations, aiming to assist clinicians with automated tools for follicle development monitoring. We propose the Weakly Supervised Follicle Segmentation (WSFS) method, a novel one-stage weakly supervised segmentation technique model designed to enhance the ultrasound images of follicles, which incorporates a Convolutional Neural Network (CNN) backbone augmented with a Feature Pyramid Network (FPN) module for multi-scale feature representation, critical for capturing the diverse sizes and shapes of follicles. By leveraging Multiple Instance Learning (MIL), we formulated the learning process in a weakly supervised manner and developed an end-to-end trainable model that efficiently addresses the issue of annotation scarcity. Furthermore, the WSFS can be used as a prompt proposal to enhance the performance of the Segmentation Anything Model (SAM), a well-known pre-trained segmentation model utilizing few-shot learning strategies. In addition, this study introduces the Follicle Ultrasound Image Dataset (FUID), addressing the scarcity in reproductive health data and aiding future research in computer-aided diagnosis. The experimental results on both the public dataset USOVA3D and private dataset FUID showed that our method performs competitively with fully supervised methods. Our approach achieves performance with mAP of 0.957, IOU of 0.714 and Dice Score of 0.83, competitive to fully supervised methods that rely on pixel-level labeled masks, despite operating with less detailed annotations.
format Article
id doaj-art-b69b9ebcf8fe4a21816e1218ddf05bc9
institution DOAJ
issn 2045-2322
language English
publishDate 2025-04-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-b69b9ebcf8fe4a21816e1218ddf05bc92025-08-20T03:14:02ZengNature PortfolioScientific Reports2045-23222025-04-0115111310.1038/s41598-025-95957-0A weakly-supervised follicle segmentation method in ultrasound imagesGuanyu Liu0Weihong Huang1Yanping Li2Qiong Zhang3Jing Fu4Hongying Tang5Jia Huang6Zhongteng Zhang7Lei Zhang8Yu Wang9Jianzhong Hu10Big Data Institute, Central South UniversityBig Data Institute, Central South UniversityDepartment of Reproductive Medicine, Xiangya Hospital, Central South UniversityDepartment of Reproductive Medicine, Xiangya Hospital, Central South UniversityDepartment of Reproductive Medicine, Xiangya Hospital, Central South UniversityDepartment of Reproductive Medicine, Xiangya Hospital, Central South UniversitySchool of Life Science, Central South UniversitySchool of Computer Sciences and Engineering, Central South UniversityLaboratory of Vision Engineering (LoVE), School of computer science, University of LincolnDepartment of Reproductive Medicine, Xiangya Hospital, Central South UniversityBig Data Institute, Central South UniversityAbstract Accurate follicle segmentation in ultrasound images is crucial for monitoring follicle development, a key factor in fertility treatments. However, obtaining pixel-level annotations for fully supervised instance segmentation is often impractical due to time and workload constraints. This paper presents a weakly supervised instance segmentation method that leverages bounding boxes as approximate annotations, aiming to assist clinicians with automated tools for follicle development monitoring. We propose the Weakly Supervised Follicle Segmentation (WSFS) method, a novel one-stage weakly supervised segmentation technique model designed to enhance the ultrasound images of follicles, which incorporates a Convolutional Neural Network (CNN) backbone augmented with a Feature Pyramid Network (FPN) module for multi-scale feature representation, critical for capturing the diverse sizes and shapes of follicles. By leveraging Multiple Instance Learning (MIL), we formulated the learning process in a weakly supervised manner and developed an end-to-end trainable model that efficiently addresses the issue of annotation scarcity. Furthermore, the WSFS can be used as a prompt proposal to enhance the performance of the Segmentation Anything Model (SAM), a well-known pre-trained segmentation model utilizing few-shot learning strategies. In addition, this study introduces the Follicle Ultrasound Image Dataset (FUID), addressing the scarcity in reproductive health data and aiding future research in computer-aided diagnosis. The experimental results on both the public dataset USOVA3D and private dataset FUID showed that our method performs competitively with fully supervised methods. Our approach achieves performance with mAP of 0.957, IOU of 0.714 and Dice Score of 0.83, competitive to fully supervised methods that rely on pixel-level labeled masks, despite operating with less detailed annotations.https://doi.org/10.1038/s41598-025-95957-0Weakly Supervised LearningUltrasound Image SegmentationAssisted Reproductive Technology
spellingShingle Guanyu Liu
Weihong Huang
Yanping Li
Qiong Zhang
Jing Fu
Hongying Tang
Jia Huang
Zhongteng Zhang
Lei Zhang
Yu Wang
Jianzhong Hu
A weakly-supervised follicle segmentation method in ultrasound images
Scientific Reports
Weakly Supervised Learning
Ultrasound Image Segmentation
Assisted Reproductive Technology
title A weakly-supervised follicle segmentation method in ultrasound images
title_full A weakly-supervised follicle segmentation method in ultrasound images
title_fullStr A weakly-supervised follicle segmentation method in ultrasound images
title_full_unstemmed A weakly-supervised follicle segmentation method in ultrasound images
title_short A weakly-supervised follicle segmentation method in ultrasound images
title_sort weakly supervised follicle segmentation method in ultrasound images
topic Weakly Supervised Learning
Ultrasound Image Segmentation
Assisted Reproductive Technology
url https://doi.org/10.1038/s41598-025-95957-0
work_keys_str_mv AT guanyuliu aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT weihonghuang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT yanpingli aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT qiongzhang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT jingfu aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT hongyingtang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT jiahuang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT zhongtengzhang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT leizhang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT yuwang aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT jianzhonghu aweaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT guanyuliu weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT weihonghuang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT yanpingli weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT qiongzhang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT jingfu weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT hongyingtang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT jiahuang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT zhongtengzhang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT leizhang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT yuwang weaklysupervisedfolliclesegmentationmethodinultrasoundimages
AT jianzhonghu weaklysupervisedfolliclesegmentationmethodinultrasoundimages