Assessment of Radiomics and Deep Learning for the Segmentation of Fetal and Maternal Anatomy in Magnetic Resonance Imaging and Ultrasound
Abstract
Recent advances in fetal imaging open the door to enhanced detection of fetal disorders and computer-assisted surgical planning. However, precise segmentation of womb's tissues is challenging due to motion artifacts caused by fetal movements and maternal respiration during acquisition. This work aims to efficiently segment different intrauterine tissues in fetal magnetic resonance imaging (MRI) and 3D ultrasound (US). First, a large set of ninety-four radiomic features are extracted to characterize the mother uterus, placenta, umbilical cord, fetal lungs, and brain. The optimal features for each anatomy are identified using both K-best and Sequential Forward Feature Selection techniques. These features are then fed to a Support Vector Machine with instance balancing to accurately segment the intrauterine anatomies. To the best of our knowledge, this is the first time that "Radiomics" is expanded from classification tasks to segmentation purposes to deal with challenging fetal images. In addition, we evaluate several state-of-the-art deep learning-based segmentation approaches. Validation is extensively performed on a set of 60 axial MRI and 3D US images from pathological and clinical cases. Our results suggest that combining the selected 10 radiomic features per anatomy along with DeepLabV3+ or BiSeNet architectures for MRI, and PSPNet or Tiramisu for 3D US, can lead to the highest fetal / maternal tissue segmentation performance, robustness, informativeness, and heterogeneity. Therefore, this work opens new avenues for advancement of segmentation techniques and, in particular, for improved fetal surgical planning.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The model by Torrents-Barrena et al. fully simulates the intraabdominal environment and enables correct entry point planning and surgeon's training.
“This model fully simulates the intraabdominal environment and enables the correct entry point planning and surgeon’s training [127,128].”
Torrents-Barrena et al. proposed a new AI algorithm for the simulation and planning of fetoscopic surgery through the detection and mapping of maternal soft tissue, uterus, placenta, and umbilical cord via MRI in combination with the detection of the placenta and its vascular tree in 3D US.
“They proposed a new AI algorithm for the simulation and planning of fetoscopic surgery through the detection and mapping of the maternal soft tissue, uterus, placenta, and umbilical cord via MRI in combination with the detection of the placenta and its vascular tree in 3D US.”
Advantages in pre-operative risk stratification or intraoperative assistance are described in both subspecialties of OB/GYN.
“the advantages in pre-operative risk stratification or intraoperative assistance are described in both subspecialties of OB/GYN, e.g., in pre-operative endometrial cancer staging [209] and for fetoscopic surgical interventions [69,127].”
AI model performance in 2D and 3D US imaging can be limited due to imaging artefacts and noise, especially when automated tissue analysis is intended.
“AI model performance in 2D and 3D US imaging can be limited due to imaging artefacts and noise, especially when automated tissue analysis is intended, for example, for fetal lung assessment”
MRI seems future-oriented and promising for fetal lung assessment.
“whereas the method of MRI for this specific application seems future-oriented and promising [127,163].”
Some authors noted the missing possibility for real-time application of various AI models, especially those designed for intrapartum application.
“some authors noted the missing possibility for real-time application of various AI models, interestingly affecting especially those that are designed for intrapartum application [42,127,172,173,174].”
Torrents-Barrena et al. demonstrated that a combination of 10 selected radiomic features led to the highest tissue segmentation performance.
“Their results demonstrated that a combination of 10 selected radiomic features led to the highest tissue segmentation performance.”
From Deep Learning Strategies for Ultrasound in Pregnancy (2020) · cites this paper as [25]
The citation lists haven’t been fetched yet
References 63 · Cited by 42
- Authors
- Torrents-Barrena, J., Monill, N., Piella, G., Gratacós, E., Eixarch, E., Ceresa, M., & González Ballester, M. A.
- Affiliations
- Universitat Pompeu Fabra (ES)
- Hospital Sant Joan de Déu Barcelona (ES)
- ORCID
- Venue
- Academic Radiology
- Publisher
- Elsevier BV
- Published
- 2021-02
- Volume
- 28
- Issue
- 2
- Pages
- 173-188
- ISSN
- 1076-6332 (ISSN-L)
- OpenAlex
- W2998129024
- Cited by
- 42
- References
- 63
- Citations by year
- 2026: 3 · 2025: 8 · 2024: 7 · 2023: 9 · 2022: 8 · 2021: 5 · 2020: 2
- FWCI
- 1.68
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Academic Radiology (publishedVersion)
- PubMed (publishedVersion)
- Funders
- Fundación Cellex
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Medical Image Segmentation Techniques
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- radiomics, placenta segmentation, fetal brain segmentation, fetal MRI, 3D ultrasound, feature selection, support vector machine, DeepLabV3+, BiSeNet, PSPNet, tiramisu
- MeSH terms
- Deep Learning, Female, Humans, Magnetic Resonance Imaging, Pregnancy, Prenatal Diagnosis, Ultrasonography, Imaging, Three-Dimensional
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
