Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging.
<h4>Purpose</h4>Accurate liver segmentation is key for volumetry assessment to guide treatment decisions. Moreover, it is an important pre-processing step for cancer detection algorithms. Liver segmentation can be especially challenging in patients with cancer-related tissue changes and...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2021-01-01
|
| Series: | PLoS ONE |
| Online Access: | https://doi.org/10.1371/journal.pone.0260630 |
| _version_ | 1828124574754013184 |
|---|---|
| author | Moritz Gross Michael Spektor Ariel Jaffe Ahmet S Kucukkaya Simon Iseke Stefan P Haider Mario Strazzabosco Julius Chapiro John A Onofrey |
| author_facet | Moritz Gross Michael Spektor Ariel Jaffe Ahmet S Kucukkaya Simon Iseke Stefan P Haider Mario Strazzabosco Julius Chapiro John A Onofrey |
| author_sort | Moritz Gross |
| collection | DOAJ |
| description | <h4>Purpose</h4>Accurate liver segmentation is key for volumetry assessment to guide treatment decisions. Moreover, it is an important pre-processing step for cancer detection algorithms. Liver segmentation can be especially challenging in patients with cancer-related tissue changes and shape deformation. The aim of this study was to assess the ability of state-of-the-art deep learning 3D liver segmentation algorithms to generalize across all different Barcelona Clinic Liver Cancer (BCLC) liver cancer stages.<h4>Methods</h4>This retrospective study, included patients from an institutional database that had arterial-phase T1-weighted magnetic resonance images with corresponding manual liver segmentations. The data was split into 70/15/15% for training/validation/testing each proportionally equal across BCLC stages. Two 3D convolutional neural networks were trained using identical U-net-derived architectures with equal sized training datasets: one spanning all BCLC stages ("All-Stage-Net": AS-Net), and one limited to early and intermediate BCLC stages ("Early-Intermediate-Stage-Net": EIS-Net). Segmentation accuracy was evaluated by the Dice Similarity Coefficient (DSC) on a dataset spanning all BCLC stages and a Wilcoxon signed-rank test was used for pairwise comparisons.<h4>Results</h4>219 subjects met the inclusion criteria (170 males, 49 females, 62.8±9.1 years) from all BCLC stages. Both networks were trained using 129 subjects: AS-Net training comprised 19, 74, 18, 8, and 10 BCLC 0, A, B, C, and D patients, respectively; EIS-Net training comprised 21, 86, and 22 BCLC 0, A, and B patients, respectively. DSCs (mean±SD) were 0.954±0.018 and 0.946±0.032 for AS-Net and EIS-Net (p<0.001), respectively. The AS-Net 0.956±0.014 significantly outperformed the EIS-Net 0.941±0.038 on advanced BCLC stages (p<0.001) and yielded similarly good segmentation performance on early and intermediate stages (AS-Net: 0.952±0.021; EIS-Net: 0.949±0.027; p = 0.107).<h4>Conclusion</h4>To ensure robust segmentation performance across cancer stages that is independent of liver shape deformation and tumor burden, it is critical to train deep learning models on heterogeneous imaging data spanning all BCLC stages. |
| first_indexed | 2024-04-11T15:07:53Z |
| format | Article |
| id | doaj.art-01bb6031f93046189a60d9ae08380ebb |
| institution | Directory Open Access Journal |
| issn | 1932-6203 |
| language | English |
| last_indexed | 2024-04-11T15:07:53Z |
| publishDate | 2021-01-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLoS ONE |
| spelling | doaj.art-01bb6031f93046189a60d9ae08380ebb2022-12-22T04:16:44ZengPublic Library of Science (PLoS)PLoS ONE1932-62032021-01-011612e026063010.1371/journal.pone.0260630Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging.Moritz GrossMichael SpektorAriel JaffeAhmet S KucukkayaSimon IsekeStefan P HaiderMario StrazzaboscoJulius ChapiroJohn A Onofrey<h4>Purpose</h4>Accurate liver segmentation is key for volumetry assessment to guide treatment decisions. Moreover, it is an important pre-processing step for cancer detection algorithms. Liver segmentation can be especially challenging in patients with cancer-related tissue changes and shape deformation. The aim of this study was to assess the ability of state-of-the-art deep learning 3D liver segmentation algorithms to generalize across all different Barcelona Clinic Liver Cancer (BCLC) liver cancer stages.<h4>Methods</h4>This retrospective study, included patients from an institutional database that had arterial-phase T1-weighted magnetic resonance images with corresponding manual liver segmentations. The data was split into 70/15/15% for training/validation/testing each proportionally equal across BCLC stages. Two 3D convolutional neural networks were trained using identical U-net-derived architectures with equal sized training datasets: one spanning all BCLC stages ("All-Stage-Net": AS-Net), and one limited to early and intermediate BCLC stages ("Early-Intermediate-Stage-Net": EIS-Net). Segmentation accuracy was evaluated by the Dice Similarity Coefficient (DSC) on a dataset spanning all BCLC stages and a Wilcoxon signed-rank test was used for pairwise comparisons.<h4>Results</h4>219 subjects met the inclusion criteria (170 males, 49 females, 62.8±9.1 years) from all BCLC stages. Both networks were trained using 129 subjects: AS-Net training comprised 19, 74, 18, 8, and 10 BCLC 0, A, B, C, and D patients, respectively; EIS-Net training comprised 21, 86, and 22 BCLC 0, A, and B patients, respectively. DSCs (mean±SD) were 0.954±0.018 and 0.946±0.032 for AS-Net and EIS-Net (p<0.001), respectively. The AS-Net 0.956±0.014 significantly outperformed the EIS-Net 0.941±0.038 on advanced BCLC stages (p<0.001) and yielded similarly good segmentation performance on early and intermediate stages (AS-Net: 0.952±0.021; EIS-Net: 0.949±0.027; p = 0.107).<h4>Conclusion</h4>To ensure robust segmentation performance across cancer stages that is independent of liver shape deformation and tumor burden, it is critical to train deep learning models on heterogeneous imaging data spanning all BCLC stages.https://doi.org/10.1371/journal.pone.0260630 |
| spellingShingle | Moritz Gross Michael Spektor Ariel Jaffe Ahmet S Kucukkaya Simon Iseke Stefan P Haider Mario Strazzabosco Julius Chapiro John A Onofrey Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging. PLoS ONE |
| title | Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging. |
| title_full | Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging. |
| title_fullStr | Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging. |
| title_full_unstemmed | Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging. |
| title_short | Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging. |
| title_sort | improved performance and consistency of deep learning 3d liver segmentation with heterogeneous cancer stages in magnetic resonance imaging |
| url | https://doi.org/10.1371/journal.pone.0260630 |
| work_keys_str_mv | AT moritzgross improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT michaelspektor improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT arieljaffe improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT ahmetskucukkaya improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT simoniseke improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT stefanphaider improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT mariostrazzabosco improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT juliuschapiro improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging AT johnaonofrey improvedperformanceandconsistencyofdeeplearning3dliversegmentationwithheterogeneouscancerstagesinmagneticresonanceimaging |