MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution

Background and objectivesThis paper introduces a novel lightweight MM-3DUNet (Multi-task Mobile 3D UNet) network designed for efficient and accurate segmentation of breast cancer tumors masses from MRI images, which leverages depth-wise separable convolutions, channel expansion units, and auxiliary...

Full description

Saved in:
Bibliographic Details
Main Authors: Xian Wang, Wenzhi Zeng, Junzeng Xu, Senhao Zhang, Yuexing Gu, Benhui Li, Xueyang Wang
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-05-01
Series:Frontiers in Oncology
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fonc.2025.1563959/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850276251820883968
author Xian Wang
Wenzhi Zeng
Junzeng Xu
Senhao Zhang
Yuexing Gu
Benhui Li
Xueyang Wang
author_facet Xian Wang
Wenzhi Zeng
Junzeng Xu
Senhao Zhang
Yuexing Gu
Benhui Li
Xueyang Wang
author_sort Xian Wang
collection DOAJ
description Background and objectivesThis paper introduces a novel lightweight MM-3DUNet (Multi-task Mobile 3D UNet) network designed for efficient and accurate segmentation of breast cancer tumors masses from MRI images, which leverages depth-wise separable convolutions, channel expansion units, and auxiliary classification tasks to enhance feature representation and computational efficiency.MethodsWe propose a 3D depth-wise separable convolution, and construct channel expansional convolution (CEC) unit and inverted residual block (IRB) to reduce the parameter count and computational load, making the network more suitable for use in resource-constrained environments. In addition, an auxiliary classification task (ACT) is introduced in the proposed architecture to provide additional supervisory signals for the main task of segmentation. The network architecture features a contracting path for downsampling and an expanding path for precise localization, enhanced by skip connections that integrate multi-level semantic information.ResultsThe network was evaluated using a dataset of Dynamic Contrast Enhanced MRI (DCE-MRI) breast cancer images, and the results show that compared to the classical 3DU-Net, MM-3DUNet could significantly reduce model parameters by 63.16% and computational demands by 80.90%, while increasing segmentation accuracy by 1.30% in IoU (Intersection over Union).ConclusionsMM-3DUNet offers a substantial reduction in computational requirements of breast cancer mass segmentation network. This network not only enhances diagnostic precision but also supports deployment in diverse clinical settings, potentially improving early detection and treatment outcomes for breast cancer patients.
format Article
id doaj-art-98528c68a9394f87a8db983e1827f53a
institution OA Journals
issn 2234-943X
language English
publishDate 2025-05-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Oncology
spelling doaj-art-98528c68a9394f87a8db983e1827f53a2025-08-20T01:50:22ZengFrontiers Media S.A.Frontiers in Oncology2234-943X2025-05-011510.3389/fonc.2025.15639591563959MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolutionXian Wang0Wenzhi Zeng1Junzeng Xu2Senhao Zhang3Yuexing Gu4Benhui Li5Xueyang Wang6Attending Physician of Health Management Institute, The Second Medical Center and National Clinical Research Center for Geriatric Diseases, Chinese PLA General Hospital, Beijing, ChinaGroup of Agricultural High-Efficiency Water Management and Artificial Intelligence, College of Agricultural Science and Engineering, Hohai University, Nanjing, Jiangsu, ChinaGroup of Agricultural High-Efficiency Water Management and Artificial Intelligence, College of Agricultural Science and Engineering, Hohai University, Nanjing, Jiangsu, ChinaDepartment of Radiology and Nuclear Medicine, Xuanwu Hospital, Capital Medical University, Beijing, ChinaDepartment of Cardiology, Yancheng Traditional Chinese Medicine Hospital Affiliated to Nanjing University of Chinese Medicine, Yancheng, Jiangsu, ChinaDepartment of Radiology, Yancheng Traditional Chinese Medicine Hospital Affiliated to Nanjing University of Chinese Medicine, Yancheng, Jiangsu, ChinaDepartment of Radiology, Yancheng Traditional Chinese Medicine Hospital Affiliated to Nanjing University of Chinese Medicine, Yancheng, Jiangsu, ChinaBackground and objectivesThis paper introduces a novel lightweight MM-3DUNet (Multi-task Mobile 3D UNet) network designed for efficient and accurate segmentation of breast cancer tumors masses from MRI images, which leverages depth-wise separable convolutions, channel expansion units, and auxiliary classification tasks to enhance feature representation and computational efficiency.MethodsWe propose a 3D depth-wise separable convolution, and construct channel expansional convolution (CEC) unit and inverted residual block (IRB) to reduce the parameter count and computational load, making the network more suitable for use in resource-constrained environments. In addition, an auxiliary classification task (ACT) is introduced in the proposed architecture to provide additional supervisory signals for the main task of segmentation. The network architecture features a contracting path for downsampling and an expanding path for precise localization, enhanced by skip connections that integrate multi-level semantic information.ResultsThe network was evaluated using a dataset of Dynamic Contrast Enhanced MRI (DCE-MRI) breast cancer images, and the results show that compared to the classical 3DU-Net, MM-3DUNet could significantly reduce model parameters by 63.16% and computational demands by 80.90%, while increasing segmentation accuracy by 1.30% in IoU (Intersection over Union).ConclusionsMM-3DUNet offers a substantial reduction in computational requirements of breast cancer mass segmentation network. This network not only enhances diagnostic precision but also supports deployment in diverse clinical settings, potentially improving early detection and treatment outcomes for breast cancer patients.https://www.frontiersin.org/articles/10.3389/fonc.2025.1563959/fullmulti-task mobile 3D UNetdynamic contrast enhanced MRIbreast cancer images segmentationresource-constrained environmentsconvolutional neural networks
spellingShingle Xian Wang
Wenzhi Zeng
Junzeng Xu
Senhao Zhang
Yuexing Gu
Benhui Li
Xueyang Wang
MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution
Frontiers in Oncology
multi-task mobile 3D UNet
dynamic contrast enhanced MRI
breast cancer images segmentation
resource-constrained environments
convolutional neural networks
title MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution
title_full MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution
title_fullStr MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution
title_full_unstemmed MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution
title_short MM-3D Unet: development of a lightweight breast cancer tumor segmentation network utilizing multi-task and depthwise separable convolution
title_sort mm 3d unet development of a lightweight breast cancer tumor segmentation network utilizing multi task and depthwise separable convolution
topic multi-task mobile 3D UNet
dynamic contrast enhanced MRI
breast cancer images segmentation
resource-constrained environments
convolutional neural networks
url https://www.frontiersin.org/articles/10.3389/fonc.2025.1563959/full
work_keys_str_mv AT xianwang mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution
AT wenzhizeng mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution
AT junzengxu mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution
AT senhaozhang mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution
AT yuexinggu mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution
AT benhuili mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution
AT xueyangwang mm3dunetdevelopmentofalightweightbreastcancertumorsegmentationnetworkutilizingmultitaskanddepthwiseseparableconvolution