Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images
Abstract Preterm birth (PTB) is the leading cause of perinatal death, affecting 10% of pregnancies. Currently, transvaginal ultrasound (TVUS) measurement of cervical length (CL) is the sole quantitative imaging metric for PTB risk, but offers limited predictive value. While computational models of c...
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-05-01
|
| Series: | npj Women's Health |
| Online Access: | https://doi.org/10.1038/s44294-025-00075-x |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849729023497732096 |
|---|---|
| author | Alicia B. Dagle Yucheng Liu Madeline Skeel Gabriel G. Trigo David Crosby Helen Feltovich Michael House Qi Yan Kristin M. Myers Sachin Jambawalikar |
| author_facet | Alicia B. Dagle Yucheng Liu Madeline Skeel Gabriel G. Trigo David Crosby Helen Feltovich Michael House Qi Yan Kristin M. Myers Sachin Jambawalikar |
| author_sort | Alicia B. Dagle |
| collection | DOAJ |
| description | Abstract Preterm birth (PTB) is the leading cause of perinatal death, affecting 10% of pregnancies. Currently, transvaginal ultrasound (TVUS) measurement of cervical length (CL) is the sole quantitative imaging metric for PTB risk, but offers limited predictive value. While computational models of cervical biomechanics show promise as PTB risk predictors, they require precise clinician-provided measurements. AI-enabled ultrasound segmentation offers a solution by automatically extracting anatomical features, thus addressing the labeling bottleneck. This study utilizes an ensemble of deep learning-based multi-class segmentation models trained on diverse TVUS data (N = 246) and evaluated on an out-of-distribution dataset (N = 29). High agreement (Dice metric ~ 0.8) between expert and model labels demonstrates the utility of AI tools in accurately measuring cervical geometry. Ultimately, this can enhance biomechanical models and more sophisticated AI-based models to better predict birth timing, specifically targeting PTB risk. |
| format | Article |
| id | doaj-art-e9bed60bece44f18b8a5f8c1bf6e244a |
| institution | DOAJ |
| issn | 2948-1716 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Women's Health |
| spelling | doaj-art-e9bed60bece44f18b8a5f8c1bf6e244a2025-08-20T03:09:20ZengNature Portfolionpj Women's Health2948-17162025-05-013111110.1038/s44294-025-00075-xGenerating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound imagesAlicia B. Dagle0Yucheng Liu1Madeline Skeel2Gabriel G. Trigo3David Crosby4Helen Feltovich5Michael House6Qi Yan7Kristin M. Myers8Sachin Jambawalikar9Department of Mechanical Engineering, Columbia UniversityDepartment of Medical Physics, Atlantic Health SystemDepartment of Computer Science, Columbia UniversityDepartment of Computer Science, Columbia UniversityDepartment of Obstetrics and Gynecology, National Maternity Hospital and University CollegeDepartment of Obstetrics and Gynecology, North Memorial Health SystemDepartment of Obstetrics and Gynecology, Tufts Medical CenterDepartment of Obstetrics and Gynecology, Columbia University Irving Medical CenterDepartment of Mechanical Engineering, Columbia UniversityDepartment of Radiology, Columbia University Irving Medical CenterAbstract Preterm birth (PTB) is the leading cause of perinatal death, affecting 10% of pregnancies. Currently, transvaginal ultrasound (TVUS) measurement of cervical length (CL) is the sole quantitative imaging metric for PTB risk, but offers limited predictive value. While computational models of cervical biomechanics show promise as PTB risk predictors, they require precise clinician-provided measurements. AI-enabled ultrasound segmentation offers a solution by automatically extracting anatomical features, thus addressing the labeling bottleneck. This study utilizes an ensemble of deep learning-based multi-class segmentation models trained on diverse TVUS data (N = 246) and evaluated on an out-of-distribution dataset (N = 29). High agreement (Dice metric ~ 0.8) between expert and model labels demonstrates the utility of AI tools in accurately measuring cervical geometry. Ultimately, this can enhance biomechanical models and more sophisticated AI-based models to better predict birth timing, specifically targeting PTB risk.https://doi.org/10.1038/s44294-025-00075-x |
| spellingShingle | Alicia B. Dagle Yucheng Liu Madeline Skeel Gabriel G. Trigo David Crosby Helen Feltovich Michael House Qi Yan Kristin M. Myers Sachin Jambawalikar Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images npj Women's Health |
| title | Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images |
| title_full | Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images |
| title_fullStr | Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images |
| title_full_unstemmed | Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images |
| title_short | Generating cervical anatomy labels using a deep ensemble multi-class segmentation model applied to transvaginal ultrasound images |
| title_sort | generating cervical anatomy labels using a deep ensemble multi class segmentation model applied to transvaginal ultrasound images |
| url | https://doi.org/10.1038/s44294-025-00075-x |
| work_keys_str_mv | AT aliciabdagle generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT yuchengliu generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT madelineskeel generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT gabrielgtrigo generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT davidcrosby generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT helenfeltovich generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT michaelhouse generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT qiyan generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT kristinmmyers generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages AT sachinjambawalikar generatingcervicalanatomylabelsusingadeepensemblemulticlasssegmentationmodelappliedtotransvaginalultrasoundimages |