Sara Neves Silva
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Sara Neves Silva has a total of 93 co-authors. Here are the top co-authors by number of shared publications:
- Jana Hutter
Early Life Imaging Department School of Biomedical Engineering and Imaging Sciences, King's College London London UK
10 shared publications - Jordina Aviles Verdera
Research Department of Early Life Imaging School of Biomedical Engineering and Imaging Sciences, King's College London London UK
9 shared publications Lisa Story
Research Department of Early Life Imaging School of Biomedical Engineering and Imaging Sciences, King's College London London UK
6 shared publicationsJoseph V. Hajnal
Imaging Physics and Engineering Research Department, School of Biomedical Engineering and Imaging Sciences Kings College London London UK
5 shared publicationsKuberan Pushparajah
Research Department of Early Life Imaging School of Biomedical Engineering and Imaging Sciences, King's College London London UK
5 shared publications- Sarah McElroy
Research Department of Imaging Physics and Engineering School of Biomedical Engineering and Imaging Sciences, King's College London London UK
5 shared publications
Latest Publications
- Open Access
- 10 July 2026
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 96 (5) , 2140-2151
Real‐Time Targeted Slice Reacquisition for Motion‐Corrupted Fetal Diffusion MRI
Purpose: To develop and evaluate a real‐time framework for targeted slice‐level reacquisition in fetal diffusion MRI. Methods: A DW‐SE‐EPI sequence was modified to support independent diffusion preparation per slice. A pathology‐robust nnU‐Net segmentation network, updated from a previously published baseline (V1) using pathological fetal cases and physics‐inspired augmentation (V2), and intensity‐based artifact detection drive automated reacquisition prioritization. Retrospective validation was performed on 60 cases across three cohorts at two field strengths and prospective deployment in 14 cases at two clinical sites. Results: Segmentation V2 significantly outperformed V1 across all cohorts (), with largest gains at b0 s mm (median DSC gain 0.87 at 1.5 T). Blackout detection achieved F1 = 0.81 0.29 at T = 0.35. Prospective real‐time deployment confirmed clinical feasibility at both field strengths. Conclusion: Real‐time slice‐level fetal diffusion reacquisition is demonstrated for the first time, opening a pathway toward routine quantitative fetal diffusion MRI in research and clinical settings.
- 10 July 2026
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 96 (5) , 2140-2151
Open AccessReal‐Time Targeted Slice Reacquisition for Motion‐Corrupted Fetal Diffusion MRI
Purpose: To develop and evaluate a real‐time framework for targeted slice‐level reacquisition in fetal diffusion MRI. Methods: A DW‐SE‐EPI sequence was modified to support independent diffusion preparation per slice. A pathology‐robust nnU‐Net segmentation network, updated from a previously published baseline (V1) using pathological fetal cases and physics‐inspired augmentation (V2), and intensity‐based artifact detection drive automated reacquisition prioritization. Retrospective validation was performed on 60 cases across three cohorts at two field strengths and prospective deployment in 14 cases at two clinical sites. Results: Segmentation V2 significantly outperformed V1 across all cohorts (), with largest gains at b0 s mm (median DSC gain 0.87 at 1.5 T). Blackout detection achieved F1 = 0.81 0.29 at T = 0.35. Prospective real‐time deployment confirmed clinical feasibility at both field strengths. Conclusion: Real‐time slice‐level fetal diffusion reacquisition is demonstrated for the first time, opening a pathway toward routine quantitative fetal diffusion MRI in research and clinical settings.
- 1 July 2026
- Published by Machine Learning for Biomedical Imaging in Machine Learning for Biomedical Imaging
- Vol. 2026 (July 2026) , 482-506
Evaluating Synthetic Data Generation for Domain Generalization in Fetal Brain MRI Segmentation
Fetal brain tissue segmentation from magnetic resonance imaging (MRI) is crucial for studying neurodevelopment, but remains challenging due to data heterogeneity and limited annotations. Domain randomization (DR) has recently emerged as a promising strategy for single-source domain generalization by synthesizing training images with randomized artifacts, contrast, and resolution. In this work, we investigate how to maximize the out-of-domain (OOD) generalization of DR-based methods using FetalSynthSeg as a case study. We show that simple Gaussian mixture-based intensity modeling outperforms more complex physics-based simulations and that intensity clustering (subdividing tissue classes by intensity) substantially improves OOD robustness. Evaluated on 348 fetal subjects from four sites spanning 0.55–3T and both T1w and T2w contrasts, FetalSynthSeg reaches state-of-the-art performance on several FeTA 2024 testing datasets (80–85 Dice score) and, for the first time, offers robust segmentation on modalities other than T2w for fetal brain segmentation (80 Dice on dHCP-T1w dataset). Compared with state-of-the-art methods such as BOUNTI, nnU-Net ensemble, and the FeTA 2024 winner, FetalSynthSeg delivers comparable or superior accuracy while maintaining strong robustness across domain shifts. Our code, model weights, and Docker image ready for easy inference are available at https://hub.docker.com/r/vzalevskyi/fetalsynthseg
- 1 July 2026
- Published by Machine Learning for Biomedical Imaging in Machine Learning for Biomedical Imaging
- Vol. 2026 (July 2026) , 482-506
Evaluating Synthetic Data Generation for Domain Generalization in Fetal Brain MRI Segmentation
Fetal brain tissue segmentation from magnetic resonance imaging (MRI) is crucial for studying neurodevelopment, but remains challenging due to data heterogeneity and limited annotations. Domain randomization (DR) has recently emerged as a promising strategy for single-source domain generalization by synthesizing training images with randomized artifacts, contrast, and resolution. In this work, we investigate how to maximize the out-of-domain (OOD) generalization of DR-based methods using FetalSynthSeg as a case study. We show that simple Gaussian mixture-based intensity modeling outperforms more complex physics-based simulations and that intensity clustering (subdividing tissue classes by intensity) substantially improves OOD robustness. Evaluated on 348 fetal subjects from four sites spanning 0.55–3T and both T1w and T2w contrasts, FetalSynthSeg reaches state-of-the-art performance on several FeTA 2024 testing datasets (80–85 Dice score) and, for the first time, offers robust segmentation on modalities other than T2w for fetal brain segmentation (80 Dice on dHCP-T1w dataset). Compared with state-of-the-art methods such as BOUNTI, nnU-Net ensemble, and the FeTA 2024 winner, FetalSynthSeg delivers comparable or superior accuracy while maintaining strong robustness across domain shifts. Our code, model weights, and Docker image ready for easy inference are available at https://hub.docker.com/r/vzalevskyi/fetalsynthseg
- Open Access
- 1 April 2026
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 96 (1) , 448-459
From Offline to Inline Without Pain: A Practical Framework for Translating Offline MR Reconstructions to Inline Deployment Using the Gadgetron Platform
Purpose: To develop and validate a practical, open‐source framework to overcome common issues in inline deployment of established offline MR reconstruction, including (1) scan disruption from lengthy reconstructions, (2) limited support for multi‐scan input reconstructions, (3) needs to adapt scripts for different raw‐data formats, and (4) limited guidance and experience in retaining scanner reconstructions and applying scanner‐based post‐processing to custom‐reconstructed images. Methods: The framework builds upon the Gadgetron platform as implemented on Siemens scanners and includes: (1) a general input converter to convert Gadgetron‐used ISMRMRD format raw into a Siemens format raw structure, facilitating reuse of code; (2) an asynchronous trigger‐and‐retrieve mechanism enabling long custom reconstructions without delaying scanner processes; (3) resource‐aware scheduling for parallel execution of reconstructions; (4) integrated file management to support multi‐scan inputs; and (5) preservation of scanner‐based reconstructions and post‐processing. The framework was validated on 2 Siemens scanners for SENSE, AlignedSENSE, and NUFFT reconstructions, and in a large‐cohort study. Results: Minimum code modification for inline deployment was demonstrated, and all reconstructions were successfully executed inline without disrupting scanner workflows. Images were retrieved automatically via retrieval scans or manually via retro‐reconstruction, with scanner‐based post‐processing applied to custom outputs. Multi‐sequence reconstructions were executed using GPU‐aware scheduling, confirming feasibility for large‐scale applications. In 480 examinations, inline reconstructions were retrieved in 99% of cases without disruptions. Conclusion: The framework lowers the technical barrier to inline deployment of offline reconstructions, enabling robust, scalable, and post‐processing‐compatible integration. It is openly available with documentation and demonstration cases to support reproducibility and community adoption.
- 1 April 2026
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 96 (1) , 448-459
Open AccessFrom Offline to Inline Without Pain: A Practical Framework for Translating Offline MR Reconstructions to Inline Deployment Using the Gadgetron Platform
Purpose: To develop and validate a practical, open‐source framework to overcome common issues in inline deployment of established offline MR reconstruction, including (1) scan disruption from lengthy reconstructions, (2) limited support for multi‐scan input reconstructions, (3) needs to adapt scripts for different raw‐data formats, and (4) limited guidance and experience in retaining scanner reconstructions and applying scanner‐based post‐processing to custom‐reconstructed images. Methods: The framework builds upon the Gadgetron platform as implemented on Siemens scanners and includes: (1) a general input converter to convert Gadgetron‐used ISMRMRD format raw into a Siemens format raw structure, facilitating reuse of code; (2) an asynchronous trigger‐and‐retrieve mechanism enabling long custom reconstructions without delaying scanner processes; (3) resource‐aware scheduling for parallel execution of reconstructions; (4) integrated file management to support multi‐scan inputs; and (5) preservation of scanner‐based reconstructions and post‐processing. The framework was validated on 2 Siemens scanners for SENSE, AlignedSENSE, and NUFFT reconstructions, and in a large‐cohort study. Results: Minimum code modification for inline deployment was demonstrated, and all reconstructions were successfully executed inline without disrupting scanner workflows. Images were retrieved automatically via retrieval scans or manually via retro‐reconstruction, with scanner‐based post‐processing applied to custom outputs. Multi‐sequence reconstructions were executed using GPU‐aware scheduling, confirming feasibility for large‐scale applications. In 480 examinations, inline reconstructions were retrieved in 99% of cases without disruptions. Conclusion: The framework lowers the technical barrier to inline deployment of offline reconstructions, enabling robust, scalable, and post‐processing‐compatible integration. It is openly available with documentation and demonstration cases to support reproducibility and community adoption.
- Open Access
- 14 November 2025
- Published by Springer Nature in Pediatric Radiology
- Vol. 56 (2) , 366-383
Towards automated fetal brain biometry reporting for 3-dimensional T2-weighted 0.55-3T magnetic resonance imaging at 20-40 weeks gestational age range
Background: The detailed assessment of fetal brain maturation and development involves morphological evaluation, gyration analysis, and reliable biometric measurements. Manual measurements on conventional 2-D magnetic resonance imaging (MRI) are affected by fetal motion, and there is no clear consensus regarding definitions for brain biometric parameters and anatomical landmark placements, making consistent reference plane and slice selection challenging. Automated biometry with 3-D slice-to-volume reconstruction (SVR) has the potential to improve the reliability of derived measurements, allowing precise quantification of fetal brain development. Previous published works have primarily focused on the technical feasibility of automated fetal brain biometry methods for T2-weighted (T2W) MRI. However, none have proposed solutions for automating the reporting of biometry results, which could enhance clinical utility and support real-time integration into routine clinical workflows. Furthermore, there is no consensus on a universal fetal biometry protocol for 3-D fetal MRI. Objective: To implement and validate a fully automated biometry reporting pipeline for 3-D T2W fetal brain MRI, based on deep learning biometry measurements and computation of z -scores and centiles, by comparison to normative growth charts. Materials and methods: Automated extraction of 13 routinely reported linear fetal biometry measurements using deep learning localization of anatomical landmarks in 3-D reconstructed T2W brain images based on 3-D UNet and presentation of the results in an .html report with centile calculation. The automated biometry method was quantitatively evaluated on 90 retrospective cases against expert manual measurements. Additionally, the fully automated, end-to-end biometry reporting pipeline was prospectively evaluated on 111 cases across a wide range of gestational ages, field strengths, and scanning parameters. We also generated normal centile ranges for 19-40 weeks GA range from 406 normal control datasets. Results: The retrospective quantitative evaluation demonstrated good agreement with manual measurements, with the maximum absolute difference between automated vs. manual measurement within a 1-3-mm range. In the prospective evaluation, more than 98% of landmark placements were graded as acceptable for interpretation and measurements. The processing time of the pipeline was less than 5 min per case, with measurements and centiles available at the time of reporting. Inspection of the automated landmark placement and computed biometrics took 1-3 min per case. The generated normative growth charts demonstrate strong correlation with the trends in the previously reported works. Conclusion: Our approach is the first to develop a fully automated biometry reporting pipeline for 3-D T2-weighted fetal MRI which integrates deep learning-based measurements, centile and z -score calculation vs. normative growth charts and report generation. Graphical
- 14 November 2025
- Published by Springer Nature in Pediatric Radiology
- Vol. 56 (2) , 366-383
Open AccessTowards automated fetal brain biometry reporting for 3-dimensional T2-weighted 0.55-3T magnetic resonance imaging at 20-40 weeks gestational age range
Background: The detailed assessment of fetal brain maturation and development involves morphological evaluation, gyration analysis, and reliable biometric measurements. Manual measurements on conventional 2-D magnetic resonance imaging (MRI) are affected by fetal motion, and there is no clear consensus regarding definitions for brain biometric parameters and anatomical landmark placements, making consistent reference plane and slice selection challenging. Automated biometry with 3-D slice-to-volume reconstruction (SVR) has the potential to improve the reliability of derived measurements, allowing precise quantification of fetal brain development. Previous published works have primarily focused on the technical feasibility of automated fetal brain biometry methods for T2-weighted (T2W) MRI. However, none have proposed solutions for automating the reporting of biometry results, which could enhance clinical utility and support real-time integration into routine clinical workflows. Furthermore, there is no consensus on a universal fetal biometry protocol for 3-D fetal MRI. Objective: To implement and validate a fully automated biometry reporting pipeline for 3-D T2W fetal brain MRI, based on deep learning biometry measurements and computation of z -scores and centiles, by comparison to normative growth charts. Materials and methods: Automated extraction of 13 routinely reported linear fetal biometry measurements using deep learning localization of anatomical landmarks in 3-D reconstructed T2W brain images based on 3-D UNet and presentation of the results in an .html report with centile calculation. The automated biometry method was quantitatively evaluated on 90 retrospective cases against expert manual measurements. Additionally, the fully automated, end-to-end biometry reporting pipeline was prospectively evaluated on 111 cases across a wide range of gestational ages, field strengths, and scanning parameters. We also generated normal centile ranges for 19-40 weeks GA range from 406 normal control datasets. Results: The retrospective quantitative evaluation demonstrated good agreement with manual measurements, with the maximum absolute difference between automated vs. manual measurement within a 1-3-mm range. In the prospective evaluation, more than 98% of landmark placements were graded as acceptable for interpretation and measurements. The processing time of the pipeline was less than 5 min per case, with measurements and centiles available at the time of reporting. Inspection of the automated landmark placement and computed biometrics took 1-3 min per case. The generated normative growth charts demonstrate strong correlation with the trends in the previously reported works. Conclusion: Our approach is the first to develop a fully automated biometry reporting pipeline for 3-D T2-weighted fetal MRI which integrates deep learning-based measurements, centile and z -score calculation vs. normative growth charts and report generation. Graphical
- Open Access
- 24 September 2025
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 95 (2) , 975-986
Scanner‐based real‐time automated volumetry reporting of the fetus, amniotic fluid, placenta, and umbilical cord for fetal MRI at 0.55T
Purpose: This work aims to enable real‐time automated intra‐uterine volumetric reporting and fetal weight estimation for fetal MRI, deployed directly on the scanner. Methods: A multi‐region segmentation nnUNet was trained on 146 images of 73 fetal subjects (coronal and axial orientations) for the parcellation of the fetal head, fetal body, placenta, amniotic fluid, and umbilical cord from whole uterus bSSFP stacks. A reporting tool was then developed to integrate the segmentation outputs into an automated report, providing volumetric measurements, fetal weight estimations, and z‐score visualizations. The complete pipeline was subsequently deployed on a 0.55T MRI scanner, enabling real‐time inference and fully automated reporting in the duration of the acquisition. Results: The segmentation pipeline was quantitatively and retrospectively evaluated on 36 stacks of 18 fetal subjects and demonstrated sufficient performance for all labels, with high scores (0.98) for the fetus, placenta, and amniotic fluid, and 0.91 for the umbilical cord. The prospective evaluation of the scanner deployment step was successfully performed on 50 cases, with the regional volumetric reports available directly on the scanner. Conclusions: This work demonstrated the feasibility of multi‐regional intra‐uterine segmentation, fetal weight estimation, and automated reporting in real‐time. This study provides a robust baseline solution for the integration of fully automated scanner‐based measurements into fetal MRI reports.
- 24 September 2025
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 95 (2) , 975-986
Open AccessScanner‐based real‐time automated volumetry reporting of the fetus, amniotic fluid, placenta, and umbilical cord for fetal MRI at 0.55T
Purpose: This work aims to enable real‐time automated intra‐uterine volumetric reporting and fetal weight estimation for fetal MRI, deployed directly on the scanner. Methods: A multi‐region segmentation nnUNet was trained on 146 images of 73 fetal subjects (coronal and axial orientations) for the parcellation of the fetal head, fetal body, placenta, amniotic fluid, and umbilical cord from whole uterus bSSFP stacks. A reporting tool was then developed to integrate the segmentation outputs into an automated report, providing volumetric measurements, fetal weight estimations, and z‐score visualizations. The complete pipeline was subsequently deployed on a 0.55T MRI scanner, enabling real‐time inference and fully automated reporting in the duration of the acquisition. Results: The segmentation pipeline was quantitatively and retrospectively evaluated on 36 stacks of 18 fetal subjects and demonstrated sufficient performance for all labels, with high scores (0.98) for the fetus, placenta, and amniotic fluid, and 0.91 for the umbilical cord. The prospective evaluation of the scanner deployment step was successfully performed on 50 cases, with the regional volumetric reports available directly on the scanner. Conclusions: This work demonstrated the feasibility of multi‐regional intra‐uterine segmentation, fetal weight estimation, and automated reporting in real‐time. This study provides a robust baseline solution for the integration of fully automated scanner‐based measurements into fetal MRI reports.
- Open Access
- 14 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 44 (11) , 4171-4180
HERON: High-Efficiency Real-Time Motion Quantification and Re-Acquisition for Fetal Diffusion MRI
Fetal diffusion MRI (dMRI) provides fascinating and clinically crucial insights into the microstructure of the human brain during development, but is highly sensitive to motion artifacts of fetal movement and maternal breathing, which impact data quality and limit diagnostic accuracy. This study introduces HERON, a robust, real-time, automatic pipeline designed to enhance fetal brain dMRI by performing motion assessment and re-acquisition. HERON leverages AI-driven brain localization, segmentation, and motion assessment on a clinical 0.55T scanner to automatically plan, quality check, and reacquire motion-affected dMRI volumes. Remaining inter-volume motion is corrected during post-processing. Tested in 20 cases, the pipeline effectively improved image quality, reduced intra- and inter-volume motion, and enabled more reliable quantitative analysis even in challenging cases. Excellent agreement with human observers (specificity 97%, sensitivity 92%) was shown and the mean Apparent Diffusion Coefficient and Intravoxel Incoherent Motion dropped in the majority of cases after correction. Improving fetal dMRI through an automatic AI-driven pipeline enables higher diagnostic quality and thus potentially wider use in both research and clinical applications.
- 14 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 44 (11) , 4171-4180
Open AccessHERON: High-Efficiency Real-Time Motion Quantification and Re-Acquisition for Fetal Diffusion MRI
Fetal diffusion MRI (dMRI) provides fascinating and clinically crucial insights into the microstructure of the human brain during development, but is highly sensitive to motion artifacts of fetal movement and maternal breathing, which impact data quality and limit diagnostic accuracy. This study introduces HERON, a robust, real-time, automatic pipeline designed to enhance fetal brain dMRI by performing motion assessment and re-acquisition. HERON leverages AI-driven brain localization, segmentation, and motion assessment on a clinical 0.55T scanner to automatically plan, quality check, and reacquire motion-affected dMRI volumes. Remaining inter-volume motion is corrected during post-processing. Tested in 20 cases, the pipeline effectively improved image quality, reduced intra- and inter-volume motion, and enabled more reliable quantitative analysis even in challenging cases. Excellent agreement with human observers (specificity 97%, sensitivity 92%) was shown and the mean Apparent Diffusion Coefficient and Intravoxel Incoherent Motion dropped in the majority of cases after correction. Improving fetal dMRI through an automatic AI-driven pipeline enables higher diagnostic quality and thus potentially wider use in both research and clinical applications.
- Open Access
- 3 April 2025
- Published by Elsevier in Journal of Cardiovascular Magnetic Resonance
- Vol. 27 (1) , 101888
Automatic flow planning for fetal cardiovascular magnetic resonance imaging
- 3 April 2025
- Published by Elsevier in Journal of Cardiovascular Magnetic Resonance
- Vol. 27 (1) , 101888
Open AccessAutomatic flow planning for fetal cardiovascular magnetic resonance imaging
- Open Access
- 10 March 2025
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 94 (2) , 615-624
Real‐time fetal brain and placental T2* mapping at 0.55T MRI
Purpose: To provide real‐time, organ‐specific quantitative information — specifically placental and fetal brain T2 * — to optimize and personalize fetal MRI examinations. Methods: A low‐latency setup enables real‐time processing, including segmentation, T2* fitting, and centile calculation. Two nnU‐Nets were trained on 2 989 fetal brains, and 540 placental datasets for automatic segmentation. Normative T2* curves over gestation were derived from 88 healthy cases. Prospective testing included 50 fetal MRI scans: A validation cohort (10 exams with three intra‐scan repetitions) and an evaluation cohort (40 participants). Validation was performed with Bland‐Altman assessments and Dice coefficients between repetitions, manual/automatic segmentations, and online/offline quantification. Results: T2* maps and centiles for the fetal brain and placenta were available in under one minute for all cases. The validation cohort showed robust reproducibility, with intra‐scan mean T2* differences of 1.04, −3.17, and 5.07 ms for the fetal brain and −3.15, 4.74, and 2.45 ms for the placenta. Mean T2* differences between online and offline processing were 1.63 ms and 0.16 ms for the fetal brain and placenta, respectively. Dice coefficients were for the placenta and for the fetal brain. Conclusions: Real‐time quantitative imaging supports personalized MR exams, optimizing sequence selection and working towards reducing recall rates. The ability to assess T2*, a potential biomarker for pregnancy complications, in real‐time opens new clinical possibilities. Future research will apply this pipeline to pregnancies affected by preeclampsia and growth restriction and explore MR‐guided fetal interventions.
- 10 March 2025
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 94 (2) , 615-624
Open AccessReal‐time fetal brain and placental T2* mapping at 0.55T MRI
Purpose: To provide real‐time, organ‐specific quantitative information — specifically placental and fetal brain T2 * — to optimize and personalize fetal MRI examinations. Methods: A low‐latency setup enables real‐time processing, including segmentation, T2* fitting, and centile calculation. Two nnU‐Nets were trained on 2 989 fetal brains, and 540 placental datasets for automatic segmentation. Normative T2* curves over gestation were derived from 88 healthy cases. Prospective testing included 50 fetal MRI scans: A validation cohort (10 exams with three intra‐scan repetitions) and an evaluation cohort (40 participants). Validation was performed with Bland‐Altman assessments and Dice coefficients between repetitions, manual/automatic segmentations, and online/offline quantification. Results: T2* maps and centiles for the fetal brain and placenta were available in under one minute for all cases. The validation cohort showed robust reproducibility, with intra‐scan mean T2* differences of 1.04, −3.17, and 5.07 ms for the fetal brain and −3.15, 4.74, and 2.45 ms for the placenta. Mean T2* differences between online and offline processing were 1.63 ms and 0.16 ms for the fetal brain and placenta, respectively. Dice coefficients were for the placenta and for the fetal brain. Conclusions: Real‐time quantitative imaging supports personalized MR exams, optimizing sequence selection and working towards reducing recall rates. The ability to assess T2*, a potential biomarker for pregnancy complications, in real‐time opens new clinical possibilities. Future research will apply this pipeline to pregnancies affected by preeclampsia and growth restriction and explore MR‐guided fetal interventions.
- Open Access
- 24 January 2025
- Published by Springer Nature in Pediatric Radiology
- Vol. 55 (3) , 556-569
Scanner-based real-time three-dimensional brain + body slice-to-volume reconstruction for T2-weighted 0.55-T low-field fetal magnetic resonance imaging
Background: Motion correction methods based on slice-to-volume registration (SVR) for fetal magnetic resonance imaging (MRI) allow reconstruction of three-dimensional (3-D) isotropic images of the fetal brain and body. However, all existing SVR methods are confined to research settings, which limits clinical integration. Furthermore, there have been no reported SVR solutions for low-field 0.55-T MRI. Objective: Integration of automated SVR motion correction methods directly into fetal MRI scanning process via the Gadgetron framework to enable automated T2-weighted (T2W) 3-D fetal brain and body reconstruction in the low-field 0.55-T MRI scanner within the duration of the scan. Materials and methods: A deep learning fully automated pipeline was developed for T2W 3-D rigid and deformable (D/SVR) reconstruction of the fetal brain and body of 0.55-T T2W datasets. Next, it was integrated into 0.55-T low-field MRI scanner environment via a Gadgetron workflow that enables launching of the reconstruction process directly during scanning in real-time. Results: During prospective testing on 12 cases (22–40 weeks gestational age), the fetal brain and body reconstructions were available on average 6:42 ± 3:13 min after the acquisition of the final stack and could be assessed and archived on the scanner console during the ongoing fetal MRI scan. The output image data quality was rated as good to acceptable for interpretation. The retrospective testing of the pipeline on 83 0.55-T datasets demonstrated stable reconstruction quality for low-field MRI. Conclusion: The proposed pipeline allows scanner-based prospective T2W 3-D motion correction for low-field 0.55-T fetal MRI via direct online integration into the scanner environment. Graphical
- 24 January 2025
- Published by Springer Nature in Pediatric Radiology
- Vol. 55 (3) , 556-569
Open AccessScanner-based real-time three-dimensional brain + body slice-to-volume reconstruction for T2-weighted 0.55-T low-field fetal magnetic resonance imaging
Background: Motion correction methods based on slice-to-volume registration (SVR) for fetal magnetic resonance imaging (MRI) allow reconstruction of three-dimensional (3-D) isotropic images of the fetal brain and body. However, all existing SVR methods are confined to research settings, which limits clinical integration. Furthermore, there have been no reported SVR solutions for low-field 0.55-T MRI. Objective: Integration of automated SVR motion correction methods directly into fetal MRI scanning process via the Gadgetron framework to enable automated T2-weighted (T2W) 3-D fetal brain and body reconstruction in the low-field 0.55-T MRI scanner within the duration of the scan. Materials and methods: A deep learning fully automated pipeline was developed for T2W 3-D rigid and deformable (D/SVR) reconstruction of the fetal brain and body of 0.55-T T2W datasets. Next, it was integrated into 0.55-T low-field MRI scanner environment via a Gadgetron workflow that enables launching of the reconstruction process directly during scanning in real-time. Results: During prospective testing on 12 cases (22–40 weeks gestational age), the fetal brain and body reconstructions were available on average 6:42 ± 3:13 min after the acquisition of the final stack and could be assessed and archived on the scanner console during the ongoing fetal MRI scan. The output image data quality was rated as good to acceptable for interpretation. The retrospective testing of the pipeline on 83 0.55-T datasets demonstrated stable reconstruction quality for low-field MRI. Conclusion: The proposed pipeline allows scanner-based prospective T2W 3-D motion correction for low-field 0.55-T fetal MRI via direct online integration into the scanner environment. Graphical
- Open Access
- 22 April 2024
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 92 (3) , 1263-1276
Fully automated planning for anatomical fetal brain MRI on 0.55T
Purpose: Widening the availability of fetal MRI with fully automatic real‐time planning of radiological brain planes on 0.55T MRI.Methods: Deep learning‐based detection of key brain landmarks on a whole‐uterus echo planar imaging scan enables the subsequent fully automatic planning of the radiological single‐shot Turbo Spin Echo acquisitions. The landmark detection pipeline was trained on over 120 datasets from varying field strength, echo times, and resolutions and quantitatively evaluated. The entire automatic planning solution was tested prospectively in nine fetal subjects between 20 and 37 weeks. A comprehensive evaluation of all steps, the distance between manual and automatic landmarks, the planning quality, and the resulting image quality was conducted.Results: Prospective automatic planning was performed in real‐time without latency in all subjects. The landmark detection accuracy was 4.2 2.6 mm for the fetal eyes and 6.5 3.2 for the cerebellum, planning quality was 2.4/3 (compared to 2.6/3 for manual planning) and diagnostic image quality was 2.2 compared to 2.1 for manual planning.Conclusions: Real‐time automatic planning of all three key fetal brain planes was successfully achieved and will pave the way toward simplifying the acquisition of fetal MRI thereby widening the availability of this modality in nonspecialist centers.
- 22 April 2024
- Published by Wiley in Magnetic Resonance in Medicine
- Vol. 92 (3) , 1263-1276
Open AccessFully automated planning for anatomical fetal brain MRI on 0.55T
Purpose: Widening the availability of fetal MRI with fully automatic real‐time planning of radiological brain planes on 0.55T MRI.Methods: Deep learning‐based detection of key brain landmarks on a whole‐uterus echo planar imaging scan enables the subsequent fully automatic planning of the radiological single‐shot Turbo Spin Echo acquisitions. The landmark detection pipeline was trained on over 120 datasets from varying field strength, echo times, and resolutions and quantitatively evaluated. The entire automatic planning solution was tested prospectively in nine fetal subjects between 20 and 37 weeks. A comprehensive evaluation of all steps, the distance between manual and automatic landmarks, the planning quality, and the resulting image quality was conducted.Results: Prospective automatic planning was performed in real‐time without latency in all subjects. The landmark detection accuracy was 4.2 2.6 mm for the fetal eyes and 6.5 3.2 for the cerebellum, planning quality was 2.4/3 (compared to 2.6/3 for manual planning) and diagnostic image quality was 2.2 compared to 2.1 for manual planning.Conclusions: Real‐time automatic planning of all three key fetal brain planes was successfully achieved and will pave the way toward simplifying the acquisition of fetal MRI thereby widening the availability of this modality in nonspecialist centers.