Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer
Uterine cervical cancer (CC) is the most common gynecologic malignancy worldwide. Whole-volume radiomic profiling from pelvic MRI may yield prognostic markers for tailoring treatment in CC. However, radiomic profiling relies on manual tumor segmentation which is unfeasible in the clinic. We present...
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-05-01
|
| Series: | Cancers |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-6694/14/10/2372 |
| _version_ | 1797501054650155008 |
|---|---|
| author | Erlend Hodneland Satheshkumar Kaliyugarasan Kari Strøno Wagner-Larsen Njål Lura Erling Andersen Hauke Bartsch Noeska Smit Mari Kyllesø Halle Camilla Krakstad Alexander Selvikvåg Lundervold Ingfrid Salvesen Haldorsen |
| author_facet | Erlend Hodneland Satheshkumar Kaliyugarasan Kari Strøno Wagner-Larsen Njål Lura Erling Andersen Hauke Bartsch Noeska Smit Mari Kyllesø Halle Camilla Krakstad Alexander Selvikvåg Lundervold Ingfrid Salvesen Haldorsen |
| author_sort | Erlend Hodneland |
| collection | DOAJ |
| description | Uterine cervical cancer (CC) is the most common gynecologic malignancy worldwide. Whole-volume radiomic profiling from pelvic MRI may yield prognostic markers for tailoring treatment in CC. However, radiomic profiling relies on manual tumor segmentation which is unfeasible in the clinic. We present a fully automatic method for the 3D segmentation of primary CC lesions using state-of-the-art deep learning (DL) techniques. In 131 CC patients, the primary tumor was manually segmented on T2-weighted MRI by two radiologists (R1, R2). Patients were separated into a train/validation (<i>n</i> = 105) and a test- (<i>n</i> = 26) cohort. The segmentation performance of the DL algorithm compared with R1/R2 was assessed with Dice coefficients (DSCs) and Hausdorff distances (HDs) in the test cohort. The trained DL network retrieved whole-volume tumor segmentations yielding median DSCs of 0.60 and 0.58 for DL compared with R1 (DL-R1) and R2 (DL-R2), respectively, whereas DSC for R1-R2 was 0.78. Agreement for primary tumor volumes was excellent between raters (R1-R2: intraclass correlation coefficient (ICC) = 0.93), but lower for the DL algorithm and the raters (DL-R1: ICC = 0.43; DL-R2: ICC = 0.44). The developed DL algorithm enables the automated estimation of tumor size and primary CC tumor segmentation. However, segmentation agreement between raters is better than that between DL algorithm and raters. |
| first_indexed | 2024-03-10T03:12:49Z |
| format | Article |
| id | doaj.art-981a29adc4d64a40b9b159ffaf1c76e9 |
| institution | Directory Open Access Journal |
| issn | 2072-6694 |
| language | English |
| last_indexed | 2024-03-10T03:12:49Z |
| publishDate | 2022-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Cancers |
| spelling | doaj.art-981a29adc4d64a40b9b159ffaf1c76e92023-11-23T10:22:14ZengMDPI AGCancers2072-66942022-05-011410237210.3390/cancers14102372Fully Automatic Whole-Volume Tumor Segmentation in Cervical CancerErlend Hodneland0Satheshkumar Kaliyugarasan1Kari Strøno Wagner-Larsen2Njål Lura3Erling Andersen4Hauke Bartsch5Noeska Smit6Mari Kyllesø Halle7Camilla Krakstad8Alexander Selvikvåg Lundervold9Ingfrid Salvesen Haldorsen10Mohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayDepartment of Obstetrics and Gynecology, Haukeland University Hospital, 5053 Bergen, NorwayDepartment of Obstetrics and Gynecology, Haukeland University Hospital, 5053 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayMohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, 5009 Bergen, NorwayUterine cervical cancer (CC) is the most common gynecologic malignancy worldwide. Whole-volume radiomic profiling from pelvic MRI may yield prognostic markers for tailoring treatment in CC. However, radiomic profiling relies on manual tumor segmentation which is unfeasible in the clinic. We present a fully automatic method for the 3D segmentation of primary CC lesions using state-of-the-art deep learning (DL) techniques. In 131 CC patients, the primary tumor was manually segmented on T2-weighted MRI by two radiologists (R1, R2). Patients were separated into a train/validation (<i>n</i> = 105) and a test- (<i>n</i> = 26) cohort. The segmentation performance of the DL algorithm compared with R1/R2 was assessed with Dice coefficients (DSCs) and Hausdorff distances (HDs) in the test cohort. The trained DL network retrieved whole-volume tumor segmentations yielding median DSCs of 0.60 and 0.58 for DL compared with R1 (DL-R1) and R2 (DL-R2), respectively, whereas DSC for R1-R2 was 0.78. Agreement for primary tumor volumes was excellent between raters (R1-R2: intraclass correlation coefficient (ICC) = 0.93), but lower for the DL algorithm and the raters (DL-R1: ICC = 0.43; DL-R2: ICC = 0.44). The developed DL algorithm enables the automated estimation of tumor size and primary CC tumor segmentation. However, segmentation agreement between raters is better than that between DL algorithm and raters.https://www.mdpi.com/2072-6694/14/10/2372cervical cancerdeep learningtumor segmentation |
| spellingShingle | Erlend Hodneland Satheshkumar Kaliyugarasan Kari Strøno Wagner-Larsen Njål Lura Erling Andersen Hauke Bartsch Noeska Smit Mari Kyllesø Halle Camilla Krakstad Alexander Selvikvåg Lundervold Ingfrid Salvesen Haldorsen Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer Cancers cervical cancer deep learning tumor segmentation |
| title | Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer |
| title_full | Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer |
| title_fullStr | Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer |
| title_full_unstemmed | Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer |
| title_short | Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer |
| title_sort | fully automatic whole volume tumor segmentation in cervical cancer |
| topic | cervical cancer deep learning tumor segmentation |
| url | https://www.mdpi.com/2072-6694/14/10/2372 |
| work_keys_str_mv | AT erlendhodneland fullyautomaticwholevolumetumorsegmentationincervicalcancer AT satheshkumarkaliyugarasan fullyautomaticwholevolumetumorsegmentationincervicalcancer AT karistrønowagnerlarsen fullyautomaticwholevolumetumorsegmentationincervicalcancer AT njallura fullyautomaticwholevolumetumorsegmentationincervicalcancer AT erlingandersen fullyautomaticwholevolumetumorsegmentationincervicalcancer AT haukebartsch fullyautomaticwholevolumetumorsegmentationincervicalcancer AT noeskasmit fullyautomaticwholevolumetumorsegmentationincervicalcancer AT marikyllesøhalle fullyautomaticwholevolumetumorsegmentationincervicalcancer AT camillakrakstad fullyautomaticwholevolumetumorsegmentationincervicalcancer AT alexanderselvikvaglundervold fullyautomaticwholevolumetumorsegmentationincervicalcancer AT ingfridsalvesenhaldorsen fullyautomaticwholevolumetumorsegmentationincervicalcancer |