Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI
Abstract Objectives Development of automated segmentation models enabling standardized volumetric quantification of fibroglandular tissue (FGT) from native volumes and background parenchymal enhancement (BPE) from subtraction volumes of dynamic contrast-enhanced breast MRI. Subsequent assessment of...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2023-11-01
|
| Series: | Insights into Imaging |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s13244-023-01531-5 |
| _version_ | 1797630188503171072 |
|---|---|
| author | Sylwia Nowakowska Karol Borkowski Carlotta M. Ruppert Anna Landsmann Magda Marcon Nicole Berger Andreas Boss Alexander Ciritsis Cristina Rossi |
| author_facet | Sylwia Nowakowska Karol Borkowski Carlotta M. Ruppert Anna Landsmann Magda Marcon Nicole Berger Andreas Boss Alexander Ciritsis Cristina Rossi |
| author_sort | Sylwia Nowakowska |
| collection | DOAJ |
| description | Abstract Objectives Development of automated segmentation models enabling standardized volumetric quantification of fibroglandular tissue (FGT) from native volumes and background parenchymal enhancement (BPE) from subtraction volumes of dynamic contrast-enhanced breast MRI. Subsequent assessment of the developed models in the context of FGT and BPE Breast Imaging Reporting and Data System (BI-RADS)-compliant classification. Methods For the training and validation of attention U-Net models, data coming from a single 3.0-T scanner was used. For testing, additional data from 1.5-T scanner and data acquired in a different institution with a 3.0-T scanner was utilized. The developed models were used to quantify the amount of FGT and BPE in 80 DCE-MRI examinations, and a correlation between these volumetric measures and the classes assigned by radiologists was performed. Results To assess the model performance using application-relevant metrics, the correlation between the volumes of breast, FGT, and BPE calculated from ground truth masks and predicted masks was checked. Pearson correlation coefficients ranging from 0.963 ± 0.004 to 0.999 ± 0.001 were achieved. The Spearman correlation coefficient for the quantitative and qualitative assessment, i.e., classification by radiologist, of FGT amounted to 0.70 (p < 0.0001), whereas BPE amounted to 0.37 (p = 0.0006). Conclusions Generalizable algorithms for FGT and BPE segmentation were developed and tested. Our results suggest that when assessing FGT, it is sufficient to use volumetric measures alone. However, for the evaluation of BPE, additional models considering voxels’ intensity distribution and morphology are required. Critical relevance statement A standardized assessment of FGT density can rely on volumetric measures, whereas in the case of BPE, the volumetric measures constitute, along with voxels’ intensity distribution and morphology, an important factor. Key points • Our work contributes to the standardization of FGT and BPE assessment. • Attention U-Net can reliably segment intricately shaped FGT and BPE structures. • The developed models were robust to domain shift. Graphical Abstract |
| first_indexed | 2024-03-11T11:04:40Z |
| format | Article |
| id | doaj.art-c02b6f553ba844d3bc6936684b713aaf |
| institution | Directory Open Access Journal |
| issn | 1869-4101 |
| language | English |
| last_indexed | 2024-03-11T11:04:40Z |
| publishDate | 2023-11-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Insights into Imaging |
| spelling | doaj.art-c02b6f553ba844d3bc6936684b713aaf2023-11-12T12:19:54ZengSpringerOpenInsights into Imaging1869-41012023-11-0114111110.1186/s13244-023-01531-5Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRISylwia Nowakowska0Karol Borkowski1Carlotta M. Ruppert2Anna Landsmann3Magda Marcon4Nicole Berger5Andreas Boss6Alexander Ciritsis7Cristina Rossi8Diagnostic and interventional Radiology, University Hospital Zurich, University Zurichb-rayZ AGDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichDiagnostic and interventional Radiology, University Hospital Zurich, University ZurichAbstract Objectives Development of automated segmentation models enabling standardized volumetric quantification of fibroglandular tissue (FGT) from native volumes and background parenchymal enhancement (BPE) from subtraction volumes of dynamic contrast-enhanced breast MRI. Subsequent assessment of the developed models in the context of FGT and BPE Breast Imaging Reporting and Data System (BI-RADS)-compliant classification. Methods For the training and validation of attention U-Net models, data coming from a single 3.0-T scanner was used. For testing, additional data from 1.5-T scanner and data acquired in a different institution with a 3.0-T scanner was utilized. The developed models were used to quantify the amount of FGT and BPE in 80 DCE-MRI examinations, and a correlation between these volumetric measures and the classes assigned by radiologists was performed. Results To assess the model performance using application-relevant metrics, the correlation between the volumes of breast, FGT, and BPE calculated from ground truth masks and predicted masks was checked. Pearson correlation coefficients ranging from 0.963 ± 0.004 to 0.999 ± 0.001 were achieved. The Spearman correlation coefficient for the quantitative and qualitative assessment, i.e., classification by radiologist, of FGT amounted to 0.70 (p < 0.0001), whereas BPE amounted to 0.37 (p = 0.0006). Conclusions Generalizable algorithms for FGT and BPE segmentation were developed and tested. Our results suggest that when assessing FGT, it is sufficient to use volumetric measures alone. However, for the evaluation of BPE, additional models considering voxels’ intensity distribution and morphology are required. Critical relevance statement A standardized assessment of FGT density can rely on volumetric measures, whereas in the case of BPE, the volumetric measures constitute, along with voxels’ intensity distribution and morphology, an important factor. Key points • Our work contributes to the standardization of FGT and BPE assessment. • Attention U-Net can reliably segment intricately shaped FGT and BPE structures. • The developed models were robust to domain shift. Graphical Abstracthttps://doi.org/10.1186/s13244-023-01531-5Fibroglandular tissue segmentationBackground parenchymal enhancement segmentationDeep learningBreast MRIAssessment standardization |
| spellingShingle | Sylwia Nowakowska Karol Borkowski Carlotta M. Ruppert Anna Landsmann Magda Marcon Nicole Berger Andreas Boss Alexander Ciritsis Cristina Rossi Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI Insights into Imaging Fibroglandular tissue segmentation Background parenchymal enhancement segmentation Deep learning Breast MRI Assessment standardization |
| title | Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI |
| title_full | Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI |
| title_fullStr | Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI |
| title_full_unstemmed | Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI |
| title_short | Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI |
| title_sort | generalizable attention u net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast dce mri |
| topic | Fibroglandular tissue segmentation Background parenchymal enhancement segmentation Deep learning Breast MRI Assessment standardization |
| url | https://doi.org/10.1186/s13244-023-01531-5 |
| work_keys_str_mv | AT sylwianowakowska generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT karolborkowski generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT carlottamruppert generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT annalandsmann generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT magdamarcon generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT nicoleberger generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT andreasboss generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT alexanderciritsis generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri AT cristinarossi generalizableattentionunetforsegmentationoffibroglandulartissueandbackgroundparenchymalenhancementinbreastdcemri |