Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach
Abstract Breast cancer is a prevalent disease affecting millions of women worldwide, and early screening can significantly reduce mortality rates. Mammograms are widely used for screening, but manual readings can lead to misdiagnosis. Computer-assisted diagnosis can help physicians make faster, more...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-07-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-09775-5 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849235497333817344 |
|---|---|
| author | Yu Wang Mudassar Ali Tariq Mahmood Amjad Rehman Tanzila Saba |
| author_facet | Yu Wang Mudassar Ali Tariq Mahmood Amjad Rehman Tanzila Saba |
| author_sort | Yu Wang |
| collection | DOAJ |
| description | Abstract Breast cancer is a prevalent disease affecting millions of women worldwide, and early screening can significantly reduce mortality rates. Mammograms are widely used for screening, but manual readings can lead to misdiagnosis. Computer-assisted diagnosis can help physicians make faster, more accurate judgments, which benefits patients. However, segmenting and classifying breast masses in mammograms is challenging due to their similar shapes to the surrounding glands. Current target detection algorithms have limited applications and low accuracy. Automated segmentation of breast masses on mammograms is a significant research challenge due to its considerable classification and contouring. This study introduces the Bi-Contextual Breast Mass Segmentation Framework (Bi-CBMSegNet), a novel paradigm that enhances the precision and efficiency of breast mass segmentation within full-field mammograms. Bi-CBMSegNet employs an advanced encoder-decoder architecture comprising two distinct modules: the Global Feature Enhancement Module (GFEM) and the Local Feature Enhancement Module (LFEM). GFEM aggregates and assimilates features from all positions within the mammogram, capturing extensive contextual dependencies that facilitate the enriched representation of homogeneous regions. The LFEM module accentuates semantic information pertinent to each specific position, refining the delineation of heterogeneous regions. The efficacy of Bi-CBMSegNet has been rigorously evaluated on two publicly available mammography databases, demonstrating superior computational efficiency and performance metrics. The findings advocate for Bi-CBMSegNet to effectuate a significant leap forward in medical imaging, particularly in breast cancer screening, thereby augmenting the accuracy and efficacy of diagnostic and treatment planning processes. |
| format | Article |
| id | doaj-art-e7945ecffa0949928a46c8388c9c2b70 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-e7945ecffa0949928a46c8388c9c2b702025-08-20T04:02:45ZengNature PortfolioScientific Reports2045-23222025-07-0115112110.1038/s41598-025-09775-5Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approachYu Wang0Mudassar Ali1Tariq Mahmood2Amjad Rehman3Tanzila Saba4Shandong Research Institute of Industrial TechnologyCollege of Information Science and Electronic Engineering, Zhejiang UniversityArtificial Intelligence and Data Analytics (AIDA) Lab, CCIS Prince Sultan UniversityArtificial Intelligence and Data Analytics (AIDA) Lab, CCIS Prince Sultan UniversityArtificial Intelligence and Data Analytics (AIDA) Lab, CCIS Prince Sultan UniversityAbstract Breast cancer is a prevalent disease affecting millions of women worldwide, and early screening can significantly reduce mortality rates. Mammograms are widely used for screening, but manual readings can lead to misdiagnosis. Computer-assisted diagnosis can help physicians make faster, more accurate judgments, which benefits patients. However, segmenting and classifying breast masses in mammograms is challenging due to their similar shapes to the surrounding glands. Current target detection algorithms have limited applications and low accuracy. Automated segmentation of breast masses on mammograms is a significant research challenge due to its considerable classification and contouring. This study introduces the Bi-Contextual Breast Mass Segmentation Framework (Bi-CBMSegNet), a novel paradigm that enhances the precision and efficiency of breast mass segmentation within full-field mammograms. Bi-CBMSegNet employs an advanced encoder-decoder architecture comprising two distinct modules: the Global Feature Enhancement Module (GFEM) and the Local Feature Enhancement Module (LFEM). GFEM aggregates and assimilates features from all positions within the mammogram, capturing extensive contextual dependencies that facilitate the enriched representation of homogeneous regions. The LFEM module accentuates semantic information pertinent to each specific position, refining the delineation of heterogeneous regions. The efficacy of Bi-CBMSegNet has been rigorously evaluated on two publicly available mammography databases, demonstrating superior computational efficiency and performance metrics. The findings advocate for Bi-CBMSegNet to effectuate a significant leap forward in medical imaging, particularly in breast cancer screening, thereby augmenting the accuracy and efficacy of diagnostic and treatment planning processes.https://doi.org/10.1038/s41598-025-09775-5CancerDeep learningMass segmentationMass classificationGlobal and local feature enhancementHealth problem |
| spellingShingle | Yu Wang Mudassar Ali Tariq Mahmood Amjad Rehman Tanzila Saba Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach Scientific Reports Cancer Deep learning Mass segmentation Mass classification Global and local feature enhancement Health problem |
| title | Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach |
| title_full | Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach |
| title_fullStr | Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach |
| title_full_unstemmed | Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach |
| title_short | Robust Bi-CBMSegNet framework for advancing breast mass segmentation in mammography with a dual module encoder-decoder approach |
| title_sort | robust bi cbmsegnet framework for advancing breast mass segmentation in mammography with a dual module encoder decoder approach |
| topic | Cancer Deep learning Mass segmentation Mass classification Global and local feature enhancement Health problem |
| url | https://doi.org/10.1038/s41598-025-09775-5 |
| work_keys_str_mv | AT yuwang robustbicbmsegnetframeworkforadvancingbreastmasssegmentationinmammographywithadualmoduleencoderdecoderapproach AT mudassarali robustbicbmsegnetframeworkforadvancingbreastmasssegmentationinmammographywithadualmoduleencoderdecoderapproach AT tariqmahmood robustbicbmsegnetframeworkforadvancingbreastmasssegmentationinmammographywithadualmoduleencoderdecoderapproach AT amjadrehman robustbicbmsegnetframeworkforadvancingbreastmasssegmentationinmammographywithadualmoduleencoderdecoderapproach AT tanzilasaba robustbicbmsegnetframeworkforadvancingbreastmasssegmentationinmammographywithadualmoduleencoderdecoderapproach |