Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging
Background and purpose: Tumor delineation is required both for radiotherapy planning and quantitative imaging biomarker purposes. It is a manual, time- and labor-intensive process prone to inter- and intraobserver variations. Semi or fully automatic segmentation could provide better efficiency and c...
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-04-01
|
| Series: | Physics and Imaging in Radiation Oncology |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2405631622000410 |
| _version_ | 1818207287899062272 |
|---|---|
| author | Franziska Knuth Aurora R. Groendahl René M. Winter Turid Torheim Anne Negård Stein Harald Holmedal Kine Mari Bakke Sebastian Meltzer Cecilia M. Futsæther Kathrine R. Redalen |
| author_facet | Franziska Knuth Aurora R. Groendahl René M. Winter Turid Torheim Anne Negård Stein Harald Holmedal Kine Mari Bakke Sebastian Meltzer Cecilia M. Futsæther Kathrine R. Redalen |
| author_sort | Franziska Knuth |
| collection | DOAJ |
| description | Background and purpose: Tumor delineation is required both for radiotherapy planning and quantitative imaging biomarker purposes. It is a manual, time- and labor-intensive process prone to inter- and intraobserver variations. Semi or fully automatic segmentation could provide better efficiency and consistency. This study aimed to investigate the influence of including and combining functional with anatomical magnetic resonance imaging (MRI) sequences on the quality of automatic segmentations. Materials and methods: T2-weighted (T2w), diffusion weighted, multi-echo T2*-weighted, and contrast enhanced dynamic multi-echo (DME) MR images of eighty-one patients with rectal cancer were used in the analysis. Four classical machine learning algorithms; adaptive boosting (ADA), linear and quadratic discriminant analysis and support vector machines, were trained for automatic segmentation of tumor and normal tissue using different combinations of the MR images as input, followed by semi-automatic morphological post-processing. Manual delineations from two experts served as ground truth. The Sørensen-Dice similarity coefficient (DICE) and mean symmetric surface distance (MSD) were used as performance metric in leave-one-out cross validation. Results: Using T2w images alone, ADA outperformed the other algorithms, yielding a median per patient DICE of 0.67 and MSD of 3.6 mm. The performance improved when functional images were added and was highest for models based on either T2w and DME images (DICE: 0.72, MSD: 2.7 mm) or all four MRI sequences (DICE: 0.72, MSD: 2.5 mm). Conclusion: Machine learning models using functional MRI, in particular DME, have the potential to improve automatic segmentation of rectal cancer relative to models using T2w MRI alone. |
| first_indexed | 2024-12-12T04:26:32Z |
| format | Article |
| id | doaj.art-1518bf6f2f2041939c1275a9042a51d9 |
| institution | Directory Open Access Journal |
| issn | 2405-6316 |
| language | English |
| last_indexed | 2024-12-12T04:26:32Z |
| publishDate | 2022-04-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Physics and Imaging in Radiation Oncology |
| spelling | doaj.art-1518bf6f2f2041939c1275a9042a51d92022-12-22T00:38:11ZengElsevierPhysics and Imaging in Radiation Oncology2405-63162022-04-01227784Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imagingFranziska Knuth0Aurora R. Groendahl1René M. Winter2Turid Torheim3Anne Negård4Stein Harald Holmedal5Kine Mari Bakke6Sebastian Meltzer7Cecilia M. Futsæther8Kathrine R. Redalen9Department of Physics, Norwegian University of Science and Technology, Høgskoleringen 5, 7491 Trondheim, Norway; Corresponding author.Faculty of Science and Technology, Norwegian University of Life Sciences, Drøbakveien 31, 1432 Ås, NorwayDepartment of Physics, Norwegian University of Science and Technology, Høgskoleringen 5, 7491 Trondheim, NorwayDepartment of Informatics, University of Oslo, Gaustadalléen 23 B, 0373 Oslo, Norway; Institute for Cancer Genetics and Informatics, Oslo University Hospital, Ullernchausséen 64, 0379 Oslo, NorwayDepartment of Radiology, Akershus University Hospital, Sykehusveien 25, 1478 Nordbyhagen, Norway; Institute of Clinical Medicine, University of Oslo, Kirkeveien 166, 0450 Oslo, NorwayDepartment of Radiology, Akershus University Hospital, Sykehusveien 25, 1478 Nordbyhagen, NorwayInstitute of Clinical Medicine, University of Oslo, Kirkeveien 166, 0450 Oslo, Norway; Department of Oncology, Akershus University Hospital, Sykehusveien 25, 1478 Nordbyhagen, NorwayDepartment of Oncology, Akershus University Hospital, Sykehusveien 25, 1478 Nordbyhagen, NorwayFaculty of Science and Technology, Norwegian University of Life Sciences, Drøbakveien 31, 1432 Ås, NorwayDepartment of Physics, Norwegian University of Science and Technology, Høgskoleringen 5, 7491 Trondheim, NorwayBackground and purpose: Tumor delineation is required both for radiotherapy planning and quantitative imaging biomarker purposes. It is a manual, time- and labor-intensive process prone to inter- and intraobserver variations. Semi or fully automatic segmentation could provide better efficiency and consistency. This study aimed to investigate the influence of including and combining functional with anatomical magnetic resonance imaging (MRI) sequences on the quality of automatic segmentations. Materials and methods: T2-weighted (T2w), diffusion weighted, multi-echo T2*-weighted, and contrast enhanced dynamic multi-echo (DME) MR images of eighty-one patients with rectal cancer were used in the analysis. Four classical machine learning algorithms; adaptive boosting (ADA), linear and quadratic discriminant analysis and support vector machines, were trained for automatic segmentation of tumor and normal tissue using different combinations of the MR images as input, followed by semi-automatic morphological post-processing. Manual delineations from two experts served as ground truth. The Sørensen-Dice similarity coefficient (DICE) and mean symmetric surface distance (MSD) were used as performance metric in leave-one-out cross validation. Results: Using T2w images alone, ADA outperformed the other algorithms, yielding a median per patient DICE of 0.67 and MSD of 3.6 mm. The performance improved when functional images were added and was highest for models based on either T2w and DME images (DICE: 0.72, MSD: 2.7 mm) or all four MRI sequences (DICE: 0.72, MSD: 2.5 mm). Conclusion: Machine learning models using functional MRI, in particular DME, have the potential to improve automatic segmentation of rectal cancer relative to models using T2w MRI alone.http://www.sciencedirect.com/science/article/pii/S2405631622000410 |
| spellingShingle | Franziska Knuth Aurora R. Groendahl René M. Winter Turid Torheim Anne Negård Stein Harald Holmedal Kine Mari Bakke Sebastian Meltzer Cecilia M. Futsæther Kathrine R. Redalen Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging Physics and Imaging in Radiation Oncology |
| title | Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging |
| title_full | Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging |
| title_fullStr | Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging |
| title_full_unstemmed | Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging |
| title_short | Semi-automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging |
| title_sort | semi automatic tumor segmentation of rectal cancer based on functional magnetic resonance imaging |
| url | http://www.sciencedirect.com/science/article/pii/S2405631622000410 |
| work_keys_str_mv | AT franziskaknuth semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT aurorargroendahl semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT renemwinter semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT turidtorheim semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT annenegard semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT steinharaldholmedal semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT kinemaribakke semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT sebastianmeltzer semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT ceciliamfutsæther semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging AT kathrinerredalen semiautomatictumorsegmentationofrectalcancerbasedonfunctionalmagneticresonanceimaging |