OC07.07: Developing a transfer learning model for the prediction of Down syndrome from first trimester ultrasound images

C. Chen

DOI: https://doi.org/10.1002/uog.27768

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 19-19, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics

What problem does this paper attempt to address?

Applying machine learning in screening for Down Syndrome in both trimesters for diverse healthcare scenarios

Huy D Do,Jeroan J Allison,Hoa L Nguyen,Hai N Phung,Cuong D Tran,Giang M Le,Trang T Nguyen

DOI: https://doi.org/10.1016/j.heliyon.2024.e34476

IF: 3.776

2024-07-15

Heliyon

Abstract:Background: This paper describes the development of low-cost, effective, non-invasive machine learning-based prediction models for Down Syndrome in the first two trimesters of pregnancy in Vietnam. These models are adaptable to different situations with limited screening capacities at community-based healthcare facilities. Method: Ultrasound and biochemical testing alone and in combination, from both trimesters were employed to build prediction models based on k-Nearest Neighbor, Support Vector Machine, Random Forest, and Extreme Gradient Boosting algorithms. Results: A total of 7,076 pregnant women from a single site in Northern Vietnam were included, and 1,035 had a fetus with Down Syndrome. Combined ultrasound and biochemical testing were required to achieve the highest accuracy in trimester 2, while models based only on biochemical testing performed as well as models based on combined testing during trimester 1. In trimester 1, Extreme Gradient Boosting produced the best model with 94% accuracy and 88% AUC, while Support Vector Machine produced the best model in trimester 2 with 89% accuracy and 84% AUC. Conclusions: This study explored a range of machine learning models under different testing scenarios. Findings point to the potential feasibility of national screening, especially in settings without enough equipment and specialists, after additional model validation and fine tuning is performed.
EP07.58: Prenatal detection of Down syndrome based on second and third trimester ultrasound scan

I. Babic,A. Khashoggi,B. AlShammari

DOI: https://doi.org/10.1002/uog.28306

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 178-179, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
EP18.18: Development of a multivariate predictive model for Trisomy 21 risk assessment at 20–25 weeks of gestation

A. Sepúlveda‐Martínez,P. Socias,O. Flores,P.A. Acosta,P. Silva,R. Guiñez,H. Muñoz,S. Lopez Leiva,M. Parra‐Cordero

DOI: https://doi.org/10.1002/uog.28754

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 287-287, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
EP17.14: Prediction of stillbirths using machine learning on first trimester clinical data

O.B. Poli Netto,D. Tiezzi,F.A. Rodrigues,A. Amoafo,F. da Silva Costa,D.L. Rolnik

DOI: https://doi.org/10.1002/uog.28671

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 265-265, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
Advancing Fetal Ultrasound Diagnostics: Innovative Methodologies for Improved Accuracy in Detecting Down Syndrome

Dinesh Mavaluru,Sahithya Ravali Ravula,Jerlin Priya Lovelin Auguskani,Santhi Muttipoll Dharmarajlu,Amutha Chellathurai,Jayabrabu Ramakrishnan,Bharath Kumar Mamilla Mugaiahgari,Nadana Ravishankar

DOI: https://doi.org/10.1016/j.medengphy.2024.104132

IF: 2.356

2024-03-03

Medical Engineering & Physics

Abstract:This research work explores the integration of medical and information technology, particularly focusing on the use of data analytics and deep learning techniques in medical image processing. Specifically, it addresses the diagnosis and prediction of fetal conditions, including Down Syndrome (DS), through the analysis of ultrasound images. Despite existing methods in image segmentation, feature extraction, and classification, there is a pressing need to enhance diagnostic accuracy. Our research delves into a comprehensive literature review and presents advanced methodologies, incorporating sophisticated deep learning architectures and data augmentation techniques to improve fetal diagnosis. Moreover, the study emphasizes the clinical significance of accurate diagnostics, detailing the training and validation process of the AI model, ensuring ethical considerations, and highlighting the potential of the model in real-world clinical settings. By pushing the boundaries of current diagnostic capabilities and emphasizing rigorous clinical validation, this research work aims to contribute significantly to medical imaging and pave the way for more precise and reliable fetal health assessments.

engineering, biomedical
OP15.04: Validation of the first trimester machine learning model for predicting pre‐eclampsia in Asian populations

L. Nguyen Hoang,D.S. Sahota,R.K. Pooh,H. Duan,N. Chaiyasit,A. Sekizawa,S. Shaw,M. Choolani,S. Seshadri,P. Yapan,W. Sim,R. Ma,W. Leung,S. Lau,N. Lee,H. Leung,T. Meshali,H. Meiri,Y. Louzoun,L.C. Poon

DOI: https://doi.org/10.1002/uog.27995

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 99-99, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
OC16.03: *Improving early detection of facial clefts using first trimester sonographic markers: diagnostic accuracy based on 389,000 examinations

J. Karim,D. Di Mascio,N. Roberts,A. Papageorghiou

DOI: https://doi.org/10.1002/uog.27831

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 41-42, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
MoM Correction Model of Maternal Serum Markers in Screening the Patients with Down Syndrome at Second Trimester

ZHUO Ren-jie,QIN Xiao-an,ZHANG Tan,LIN Jun,ZHANG Li-na,CHEN Yi-zhen,HE Jing,CHEN Zhi-yang,SHEN Qi-jun

2011-01-01

Abstract:Objective To establish appropriate correction models of serum markers for prenatal screening of the patients with Down syndrome at the second trimester.Methods Nonlinear Weighted Regression models were employed to regress maternal serum markers and gestational age.The optimal models were selected by PRESS Statistics and R2adj.Then the medians were compared with those using such screening software as Lifecycle 3.0 and 2T by cumulative sum plots.Results The optimal models between maternal serum markers and gestational age were well built.The cumulative sum of log10 (MoM of AFP),log10 (MoM of free β-HCG) and log10 (MoM of uE3) for the models was near to zero,while those values for Lifecycle 3.0 and 2T were far deviated from zero.Conclusion Our correction model for maternal serum markers is more applicable than the models built by Lifecycle 3.0 and 2T for Chinese population at the second trimester screening.
EP01.02: Automatic diagnosis of Caesarean scar pregnancy based on ultrasound images by the deep learning model

G. Chen,C. Ding,K. Tang,D. Zhang,Q. Xu,L. Shao,W. Liao,Y. Wang

DOI: https://doi.org/10.1002/uog.28017

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 107-107, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
Deep learning prediction of renal anomalies for prenatal ultrasound diagnosis

Olivier X Miguel,Emily Kaczmarek,Inok Lee,Robin Ducharme,Alysha L J Dingwall-Harvey,Ruth Rennicks White,Brigitte Bonin,Richard I Aviv,Steven Hawken,Christine M Armour,Kevin Dick,Mark C Walker

DOI: https://doi.org/10.1038/s41598-024-59248-4

2024-04-19

Abstract:Deep learning algorithms have demonstrated remarkable potential in clinical diagnostics, particularly in the field of medical imaging. In this study, we investigated the application of deep learning models in early detection of fetal kidney anomalies. To provide an enhanced interpretation of those models' predictions, we proposed an adapted two-class representation and developed a multi-class model interpretation approach for problems with more than two labels and variable hierarchical grouping of labels. Additionally, we employed the explainable AI (XAI) visualization tools Grad-CAM and HiResCAM, to gain insights into model predictions and identify reasons for misclassifications. The study dataset consisted of 969 ultrasound images from unique patients; 646 control images and 323 cases of kidney anomalies, including 259 cases of unilateral urinary tract dilation and 64 cases of unilateral multicystic dysplastic kidney. The best performing model achieved a cross-validated area under the ROC curve of 91.28% ± 0.52%, with an overall accuracy of 84.03% ± 0.76%, sensitivity of 77.39% ± 1.99%, and specificity of 87.35% ± 1.28%. Our findings emphasize the potential of deep learning models in predicting kidney anomalies from limited prenatal ultrasound imagery. The proposed adaptations in model representation and interpretation represent a novel solution to multi-class prediction problems.
Down Syndrome detection with Swin Transformer architecture

Chengyu Wang,Limin Yu,Jionglong Su,Trevor Mahy,Valerio Selis,Chunxiao Yang,Fei Ma

DOI: https://doi.org/10.1016/j.bspc.2023.105199

IF: 5.1

2023-07-06

Biomedical Signal Processing and Control

Abstract:Objective: Down Syndrome, also known as Trisomy 21, is a severe genetic disease caused by an extra chromosome 21. For the detection of Trisomy 21, despite those statistical methods have been widely used for screening, karyotyping remains the gold standard and the first level of testing for diagnosis. Due to karyotyping being a time-consuming and labour-intensive procedure, Computer Vision methodologies have been explored to automate the karyotyping process for decades. However, few studies have focused on Down Syndrome detection with the Transformer technique. This study develops a Down-Syndrome-Detector (DSD) architecture based on the Transformer structure, which includes a segmentation module, an alignment module, a classification module, and a Down Syndrome indicator. Methods: The segmentation and classification modules are designed by homogeneous transfer learning at the model level. Transfer learning techniques enable a network to share weights learned from the source domain (e.g., millions of data in ImageNet) and optimize the weights with limited labeled data in the target domain (e.g., less than 6,000 images in BioImLab). The Align-Module is designed to process the segmentation output to fit the classification dataset, and the Down Syndrome Indicator identifies a Down Syndrome case from the classification output. Results: Experiments are first performed on two public datasets BioImLab (119 cases) and Advanced Digital Imaging Research (ADIR, 180 cases). Our performance metrics indicate the good ability of segmentation and classification modules of DSD . Then, the DS detection performance of DSD is evaluated on a private dataset consisting of 1084 cells (including 20 DS cells from 2 singleton cases): 90.0% and 86.1% for cell-level TPR and TNR; 100% and 96.08% for case-level TPR and TNR, respectively. Conclusion: This study develops a pipeline based on the modern Transformer architecture for the detection of Down Syndrome from original metaphase micrographs. Both segmentation and classification models developed in this study are assessed using public datasets with commonly used metrics, and both achieved good results. The DSD proposed in this study reported satisfactory singleton case-specific DS detection results. Significance: As verified by a medical specialist, the developed method may improve Down Syndrome detection efficiency by saving human labor and improving clinical practice.

engineering, biomedical
A Preliminary Study to Quantitatively Evaluate the Development of Maturation Degree for Fetal Lung Based on Transfer Learning Deep Model from Ultrasound Images

Chen Ping,Chen Yunqi,Deng Yinhui,Wang Yuanyuan,He Ping,Lv Xiaoli,Yu Jinhua

DOI: https://doi.org/10.1007/s11548-020-02211-1

2020-01-01

International Journal of Computer Assisted Radiology and Surgery

Abstract:The evaluation of fetal lung maturity is critical for clinical practice since the lung immaturity is an important cause of neonatal morbidity and mortality. For the evaluation of the development of fetal lung maturation degree, our study established a deep model from ultrasound images of four-cardiac-chamber view plane. A two-stage transfer learning approach is proposed for the purpose of the study. A specific U-net structure is designed for the applied deep model. In the first stage, the model is to first learn the recognition of fetal lung region in the ultrasound images. It is hypothesized in our study that the development of fetal lung maturation degree is generally proportional to the gestational age. Then, in the second stage, the pretrained deep model is trained to accurately estimate the gestational age from the fetal lung region of ultrasound images. Totally 332 patients were included in our study, while the first 206 patients were used for training and the subsequent 126 patients were used for the independent testing. The testing results of the established deep model have the imprecision as 1.56 ± 2.17 weeks on the gestational age estimation. Its correlation coefficient with the ground truth of gestational age achieves 0.7624 (95% CI 0.6779 to 0.8270, P value < 0.00001). The hypothesis that the development of fetal lung maturation degree can be represented by the texture information from ultrasound images has been preliminarily validated. The fetal lung maturation degree can be considered as being represented by the deep model’s output denoted by the estimated gestational age.
A multicenter study on two-stage transfer learning model for duct-dependent CHDs screening in fetal echocardiography

Jiajie Tang,Yongen Liang,Yuxuan Jiang,Jinrong Liu,Rui Zhang,Danping Huang,Chengcheng Pang,Chen Huang,Dongni Luo,Xue Zhou,Ruizhuo Li,Kanghui Zhang,Bingbing Xie,Lianting Hu,Fanfan Zhu,Huimin Xia,Long Lu,Hongying Wang

DOI: https://doi.org/10.1038/s41746-023-00883-y

IF: 15.2

2023-08-13

npj Digital Medicine

Abstract:Duct-dependent congenital heart diseases (CHDs) are a serious form of CHD with a low detection rate, especially in underdeveloped countries and areas. Although existing studies have developed models for fetal heart structure identification, there is a lack of comprehensive evaluation of the long axis of the aorta. In this study, a total of 6698 images and 48 videos are collected to develop and test a two-stage deep transfer learning model named DDCHD-DenseNet for screening critical duct-dependent CHDs. The model achieves a sensitivity of 0.973, 0.843, 0.769, and 0.759, and a specificity of 0.985, 0.967, 0.956, and 0.759, respectively, on the four multicenter test sets. It is expected to be employed as a potential automatic screening tool for hierarchical care and computer-aided diagnosis. Our two-stage strategy effectively improves the robustness of the model and can be extended to screen for other fetal heart development defects.

health care sciences & services,medical informatics
Can deep learning classify cerebral ultrasound images for the detection of brain injury in very preterm infants?

Tahani Ahmad,Alessandro Guida,Samuel Stewart,Noah Barrett,Xiang Jiang,Michael Vincer,Jehier Afifi

DOI: https://doi.org/10.1007/s00330-024-11028-4

2024-08-30

Abstract:Objectives: Cerebral ultrasound (CUS) is the main imaging screening tool in preterm infants. The aim of this work is to develop deep learning (DL) models that classify normal vs abnormal CUS to serve as a computer-aided detection tool providing timely interpretation of the scans. Methods: A population-based cohort of very preterm infants (220-306 weeks) born between 2004 and 2016 in Nova Scotia, Canada. A set of nine sequential CUS images per infant was retrieved at three specific coronal landmarks at three pre-identified times (first, sixth weeks, and term age). A radiologist manually labeled each image as normal or abnormal. The dataset was split into training/development/test subsets (80:10:10). Different convolutional neural networks were tested, with filtering of the most uncertain prediction. The model's performance was assessed using precision/recall and the receiver operating area under the curve. Results: Sequential CUS retrieved for 538/665 babies (81% of the cohort). Four thousand one hundred eighty images were used to develop and test the model. The model performance was only discrete at the beginning but, through different machine learning strategies was boosted to good levels averaging 0.86 ROC AUC (95% CI: 0.82, 0.90) and 0.87 PR AUC (95% CI: 0.84, 0.90) (model uncertainty estimation filters using normalized entropy threshold = 0.5). Conclusion: This study offers proof of the feasibility of applying DL to CUS. This basic diagnostic model showed good discriminative ability to classify normal versus abnormal CUS. This serves as a CAD and a framework for constructing a prognostic model. Clinical relevance statement: This DL model can serve as a computer-aided detection tool to classify CUS of very preterm babies as either normal or abnormal. This model will also be used as a framework to develop a prognostic model. Key points: Binary computer-aided detection models of CUS are applicable for classifying ultrasound images in very preterm babies. This model acts as a step towards developing a model for predicting neurodevelopmental outcomes in very preterm babies. This model serves as a tool for interpretation of CUS in this patient population with a heightened risk of brain injury.
OC07.03: Automated cortical analysis from third‐dimensional second trimester ultrasound scans

M.K. Wyburd,I. Consortium,M. Jenkinson,A. Namburete

DOI: https://doi.org/10.1002/uog.27764

2024-09-11

Ultrasound in Obstetrics and Gynecology

Abstract:Ultrasound in Obstetrics &Gynecology, Volume 64, Issue S1, Page 18-18, September 2024.

radiology, nuclear medicine & medical imaging,obstetrics & gynecology,acoustics
Combined measurement of fetal nuchal translucency, maternal serum free beta-hCG, and pregnancy-associated plasma protein A for first-trimester Down's syndrome screening

M S Tsai,Y Y Huang,K Y Hwa,C C Cheng,F K Lee

Abstract:Purpose: It has been proposed that first-trimester Down's syndrome screening has a higher detection rate compared to second-trimester biochemical screening. This study investigated the accuracy of Down's syndrome screening during gestational weeks 10 to 13 using the combination of fetal nuchal translucency (NT) measurement with maternal serum concentrations of free beta-human chorionic gonadotropin (beta-hCG) and pregnancy-associated plasma protein-A (PAPP-A). Methods: A total of 1,514 women with singleton pregnancies were enrolled in this study. Fetal NT was measured using the criteria published by the Fetal Medicine Foundation. Maternal serum concentrations of free beta-hCG and PAPP-A were determined by microtiter-plate ELISA. Down's syndrome risk was calculated using multivariate Gaussian distribution and Alpha software. Results: Seventeen (1.12%) of the 1514 screened pregnancies had a fetal NT of at least 3 mm, and 41.2% of these had a poor pregnancy outcome, including four fetal aneuploidies. The odds of a fetal aneuploidy when the NT was greater than 2.0 multiples of median (MoM) was 90, when serum PAPP-A concentration was less than 0.45 MoM, it was 8.6, and when serum free beta-hCG concentration was greater than 2.2 MoM, it was 4.7. Using a risk cut-off level of 1 in 400, nine of 10 fetal aneuploidies were identified with a 4.7% false-positive rate, including two with trisomy 21, one with trisomy 18, and three with Turner's syndrome. Conclusions: This study demonstrated that Down's syndrome screening using the combined test in the first trimester had a higher detection rate than that of serum screening in the second trimester. Implementation of NT measurement in the first trimester provides substantial advantages for Down's syndrome detection and early diagnosis of fetal structural abnormalities.
Screening for Down's syndrome in the first trimester: combined risk calculation, methodology, and validation of a web-based system

Marius Kublickas,Jennifer Crossley,David Aitken

DOI: https://doi.org/10.1080/00016340902859299

Abstract:Objective: To provide the necessary mathematical formulae to construct a risk calculation package for Down's syndrome using maternal serum free beta human chorionic gonadotrophin, pregnancy associated plasma protein A, and ultrasound nuchal translucency(NT) measurements in the first trimester for use in a web-based system. Setting: Swedish antenatal screening program. Population: Swedish women choosing to have first trimester screening for Down's syndrome. Methods: A standard Gaussian approach to calculation of likelihood ratios for each marker value was used following conversion of analyte concentrations and NT measurements to multiples of the normal gestational median (MoM). Validation of the algorithm was carried out on a study group of 3,900 pregnancies (including 29 cases of Down's syndrome). Predicted detection and screen positive rates were compared with an alternative calculation package. Main outcome measures: Sensitivity and specificity of screening delivered by risk calculation package. Results: A cut-off risk of 1 in 200 produced a detection rate of 90% at a 4.7% false-positive rate. In addition, use of the calculation package in a web-based system proved efficient in routine practice, providing rapid access by individual centers to results and easy central monitoring and auditing of the program. Conclusions: The system described here fulfills the requirement for an appropriately validated risk calculation package, is fully user configurable with easy access through a web-based system allowing extension of the screening program to multiple centers.
Prototypical Cross-domain Knowledge Transfer for Cervical Dysplasia Visual Inspection

Yichen Zhang,Yifang Yin,Ying Zhang,Zhenguang Liu,Zheng Wang,Roger Zimmermann

DOI: https://doi.org/10.1145/3581783.3612000

2023-08-19

Abstract:Early detection of dysplasia of the cervix is critical for cervical cancer treatment. However, automatic cervical dysplasia diagnosis via visual inspection, which is more appropriate in low-resource settings, remains a challenging problem. Though promising results have been obtained by recent deep learning models, their performance is significantly hindered by the limited scale of the available cervix datasets. Distinct from previous methods that learn from a single dataset, we propose to leverage cross-domain cervical images that were collected in different but related clinical studies to improve the model's performance on the targeted cervix dataset. To robustly learn the transferable information across datasets, we propose a novel prototype-based knowledge filtering method to estimate the transferability of cross-domain samples. We further optimize the shared feature space by aligning the cross-domain image representations simultaneously on domain level with early alignment and class level with supervised contrastive learning, which endows model training and knowledge transfer with stronger robustness. The empirical results on three real-world benchmark cervical image datasets show that our proposed method outperforms the state-of-the-art cervical dysplasia visual inspection by an absolute improvement of 4.7% in top-1 accuracy, 7.0% in precision, 1.4% in recall, 4.6% in F1 score, and 0.05 in ROC-AUC.

Computer Vision and Pattern Recognition
Novel Transfer Learning Based Deep Features for Diagnosis of Down Syndrome in Children Using Facial Images

Ali Raza,Kashif Munir,Mubarak S. Almutairi,Rukhshanda Sehar

DOI: https://doi.org/10.1109/access.2024.3359235

IF: 3.9

2024-02-07

IEEE Access

Abstract:Down syndrome is a chromosomal condition characterized by the existence of an additional copy of chromosome 21. This genetic anomaly leads to a range of developmental challenges and distinct physical characteristics in affected children. Children with Down syndrome often exhibit specific craniofacial proportions, such as a relatively shorter midface and broader facial width. These distinct facial features, including a flat nasal bridge, almond-shaped eyes, and a small and somewhat flattened head, can serve as valuable indicators for early diagnosis and intervention. This study aims at the early diagnosis of Down syndrome using an advanced neural network approach. We used 3,009 facial images of children with Down syndrome and healthy children taken from the age group range of 0 to 15 for conducting our research experiments. We proposed a novel transfer learning-based feature generation named VNL-Net, which is an ensemble of VGG16, Non-Negative Matrix Factorization (NMF), and Light Gradient Boosting Machine (LGBM) methods. This unique VNL-Net feature extraction initially extracts spatial features from input image data. Then, the ensemble feature set of NMF and LGBM is extracted from spatial features. We built several advanced artificial intelligence-based approaches on the newly created feature set to evaluate performance. Extensive research experimental results show that the logistic regression method outperformed state-of-the-art studies with a high-performance accuracy of 0.99. We also fine-tuned each applied method and validated performance using the k-fold cross-validation mechanism. The runtime computational complexity of the applied methods is also determined. Our proposed innovative research has the ability to revolutionize the early diagnosis of Down syndrome in children using facial images.

computer science, information systems,telecommunications,engineering, electrical & electronic
Deep‐learning computer vision can identify increased nuchal translucency in the first trimester of pregnancy

Bhavya Kasera,Shiri Shinar,Parinita Edke,Vagisha Pruthi,Anna Goldenberg,Lauren Erdman,Tim Van Mieghem

DOI: https://doi.org/10.1002/pd.6559

2024-04-03

Prenatal Diagnosis

Abstract:Objective Many fetal anomalies can already be diagnosed by ultrasound in the first trimester of pregnancy. Unfortunately, in clinical practice, detection rates for anomalies in early pregnancy remain low. Our aim was to use an automated image segmentation algorithm to detect one of the most common fetal anomalies: a thickened nuchal translucency (NT), which is a marker for genetic and structural anomalies. Methods Standardized mid‐sagittal ultrasound images of the fetal head and chest were collected for 560 fetuses between 11 and 13 weeks and 6 days of gestation, 88 (15.7%) of whom had an NT thicker than 3.5 mm. Image quality was graded as high or low by two fetal medicine experts. Images were divided into a training‐set (n = 451, 55 thick NT) and a test‐set (n = 109, 33 thick NT). We then trained a U‐Net convolutional neural network to segment the fetus and the NT region and computed the NT:fetus ratio of these regions. The ability of this ratio to separate thick (anomalous) NT regions from healthy, typical NT regions was first evaluated in ground‐truth segmentation to validate the metric and then with predicted segmentation to validate our algorithm, both using the area under the receiver operator curve (AUROC). Results The ground‐truth NT:fetus ratio detected thick NTs with 0.97 AUROC in both the training and test sets. The fetus and NT regions were detected with a Dice score of 0.94 in the test set. The NT:fetus ratio based on model segmentation detected thick NTs with an AUROC of 0.96 relative to clinician labels. At a 91% specificity, 94% of thick NT cases were detected (sensitivity) in the test set. The detection rate was statistically higher (p = 0.003) in high versus low‐quality images (AUROC 0.98 vs. 0.90, respectively). Conclusion Our model provides an explainable deep‐learning method for detecting increased NT. This technique can be used to screen for other fetal anomalies in the first trimester of pregnancy.

genetics & heredity,obstetrics & gynecology

OC07.07: Developing a transfer learning model for the prediction of Down syndrome from first trimester ultrasound images

Applying machine learning in screening for Down Syndrome in both trimesters for diverse healthcare scenarios

EP07.58: Prenatal detection of Down syndrome based on second and third trimester ultrasound scan

EP18.18: Development of a multivariate predictive model for Trisomy 21 risk assessment at 20–25 weeks of gestation

EP17.14: Prediction of stillbirths using machine learning on first trimester clinical data

Advancing Fetal Ultrasound Diagnostics: Innovative Methodologies for Improved Accuracy in Detecting Down Syndrome

OP15.04: Validation of the first trimester machine learning model for predicting pre‐eclampsia in Asian populations

OC16.03: *Improving early detection of facial clefts using first trimester sonographic markers: diagnostic accuracy based on 389,000 examinations

MoM Correction Model of Maternal Serum Markers in Screening the Patients with Down Syndrome at Second Trimester

EP01.02: Automatic diagnosis of Caesarean scar pregnancy based on ultrasound images by the deep learning model

Deep learning prediction of renal anomalies for prenatal ultrasound diagnosis

Down Syndrome detection with Swin Transformer architecture

A Preliminary Study to Quantitatively Evaluate the Development of Maturation Degree for Fetal Lung Based on Transfer Learning Deep Model from Ultrasound Images

A multicenter study on two-stage transfer learning model for duct-dependent CHDs screening in fetal echocardiography

Can deep learning classify cerebral ultrasound images for the detection of brain injury in very preterm infants?

OC07.03: Automated cortical analysis from third‐dimensional second trimester ultrasound scans

Combined measurement of fetal nuchal translucency, maternal serum free beta-hCG, and pregnancy-associated plasma protein A for first-trimester Down's syndrome screening

Screening for Down's syndrome in the first trimester: combined risk calculation, methodology, and validation of a web-based system

Prototypical Cross-domain Knowledge Transfer for Cervical Dysplasia Visual Inspection

Novel Transfer Learning Based Deep Features for Diagnosis of Down Syndrome in Children Using Facial Images

Deep‐learning computer vision can identify increased nuchal translucency in the first trimester of pregnancy