Dataset Information

Fetal biometry and amniotic fluid volume assessment end-to-end automation using Deep Learning.

ABSTRACT: Fetal biometry and amniotic fluid volume assessments are two essential yet repetitive tasks in fetal ultrasound screening scans, aiding in the detection of potentially life-threatening conditions. However, these assessment methods can occasionally yield unreliable results. Advances in deep learning have opened up new avenues for automated measurements in fetal ultrasound, demonstrating human-level performance in various fetal ultrasound tasks. Nevertheless, the majority of these studies are retrospective in silico studies, with a limited number including African patients in their datasets. In this study we developed and prospectively assessed the performance of deep learning models for end-to-end automation of fetal biometry and amniotic fluid volume measurements. These models were trained using a newly constructed database of 172,293 de-identified Moroccan fetal ultrasound images, supplemented with publicly available datasets. the models were then tested on prospectively acquired video clips from 172 pregnant people forming a consecutive series gathered at four healthcare centers in Morocco. Our results demonstrate that the 95% limits of agreement between the models and practitioners for the studied measurements were narrower than the reported intra- and inter-observer variability among expert human sonographers for all the parameters under study. This means that these models could be deployed in clinical conditions, to alleviate time-consuming, repetitive tasks, and make fetal ultrasound more accessible in limited-resource environments.

SUBMITTER: Slimani S

PROVIDER: S-EPMC10624828 | biostudies-literature | 2023 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Fetal biometry and amniotic fluid volume assessment end-to-end automation using Deep Learning.

Slimani Saad S Hounka Salaheddine S Mahmoudi Abdelhak A Rehah Taha T Laoudiyi Dalal D Saadi Hanane H Bouziyane Amal A Lamrissi Amine A Jalal Mohamed M Bouhya Said S Akiki Mustapha M Bouyakhf Youssef Y Badaoui Bouabid B Radgui Amina A Mhlanga Musa M Bouyakhf El Houssine EH

Nature communications 20231103 1

Fetal biometry and amniotic fluid volume assessments are two essential yet repetitive tasks in fetal ultrasound screening scans, aiding in the detection of potentially life-threatening conditions. However, these assessment methods can occasionally yield unreliable results. Advances in deep learning have opened up new avenues for automated measurements in fetal ultrasound, demonstrating human-level performance in various fetal ultrasound tasks. Nevertheless, the majority of these studies are retr ...[more]

PMID: 37923713

Similar Datasets

Project description:BackgroundAmniotic fluid is essential to normal fetal development and is estimated clinically with ultrasound scanning to identify pregnancies that are at risk for poor perinatal outcome.ObjectiveOur goal was to develop a United States standard for amniotic fluid volume that is estimated by the amniotic fluid index and single deepest pocket.Study designWe performed a planned secondary analysis of a multicenter observational study of 2334 low-risk women with normal singleton gestations from 1 of 4 self-reported racial/ethnic groups. Eligible women had confirmed first-trimester dating criteria with health status, lifestyles, and medical and obstetric histories that were associated with normal fetal growth. Consenting women underwent serial (up to 5) sonographic evaluations of amniotic fluid between 15 and 40 weeks of gestation after being assigned randomly to 1 of 4 gestational age observation schedules. Twelve United States perinatal centers participated, and all sonograms were performed by credentialed sonographers who used identical, high-resolution equipment; caregivers were unaware of results but were notified for oligohydramnios. Women (n=597) who were subsequently found to have clinically significant antepartum complications were excluded. Racial/ethnic-specific nomograms for amniotic fluid index and single deepest pocket across gestation were developed with the use of linear mixed models with cubic splines; racial/ethnic differences were evaluated both with global and between-group tests. Median, 3rd, 5th, 10th, 90th, 95th and 97th percentile values were also estimated. We further considered the possible confounding effects of selected maternal characteristics and the estimated fetal weight at each sonogram.ResultsA total of 1719 pregnant women met inclusion criteria and had available data. These included 480 non-Hispanic white women, 418 non-Hispanic black women, 485 Hispanic women, and 336 Asian women. Both the amniotic fluid index and the single deepest pocket varied across gestation with maximal values at 26 and 33 weeks of gestation, respectively. Statistically significant differences were observed by maternal race/ethnicity. The between-group differences that were observed at 17-22 and 35-40 weeks of gestation remained statistically significant after adjustment for maternal characteristics and estimated fetal weight. These between-group racial/ethnic differences were most prominent after 35 weeks of gestation and at the extremes of dispersion (3rd and 97th percentiles). All 3rd and 97th percentile amniotic fluid index values were within the range of commonly used cutoffs to define oligohydramnios (≤5 cm) and polyhydramnios (≥25 cm). However, the 3rd percentile values ranged between 5.9 cm at 40 weeks of gestation and 10.1 cm at 25-27 weeks of gestation; the 97th percentile values ranged between 24.8 cm at 38 weeks of gestation and 15.7 cm at 15 weeks of gestation.ConclusionSonographic amniotic fluid volume estimates vary by racial/ethnic group, but the absolute differences appear to be small and may not be clinically significant. Selected maternal characteristics and estimated fetal weight did not affect the racial/ethnic differences. Between-group differences are maximal after 35 weeks of gestation and at the extremes of the upper and lower dispersion estimates. Given the observed variability in extreme (3rd and 97th percentile) dispersion values over the gestation, use of single cutoffs to define out-of-range measurements may not be appropriate clinically. These data might form a contemporary United States standard for amniotic fluid estimation that uses the amniotic fluid index and the single deepest pocket.

Project description:Amniotic fluid is essential for fetus wellbeing and is used to monitor pregnancy and predict fetal outcomes. Sex affects health and medicine from the beginning of life, but knowledge of its influence on cell-depleted amniotic fluid (AF) and amniotic fluid cells (AFCs) is still neglected. We evaluated sex-related differences in AF and in AFCs to extend personalized medicine to prenatal life. AFCs and AF were obtained from healthy Caucasian pregnant women who underwent amniocentesis at the 16th-18th week of gestation for advanced maternal age. In the AF, inflammation biomarkers (TNFα, IL6, IL8, and IL4), malondialdehyde, nitrites, amino acids, and acylcarnitines were measured. Estrogen receptors and cell fate (autophagy, apoptosis, senescence) were measured in AFCs. TNFα, IL8, and IL4 were higher in female AF, whereas IL6, nitrites, and MDA were similar. Valine was higher in male AF, whereas several acylcarnitines were sexually different, suggesting a mitochondrial involvement in establishing sex differences. Female AFCs displayed higher expression of ERα protein and a higher ERα/ERβ ratio. The ratio of LC3II/I, an index of autophagy, was higher in female AFCs, while LC3 gene was similar in both sexes. No significant sex differences were found in the expression of the lysosomal protein LAMP1, while p62 was higher in male AFCs. LAMP1 gene was upregulated in male AFCs, while p62 gene was upregulated in female ones. Finally, caspase 9 activity and senescence linked to telomeres were higher in female AFCs, while caspase 3 and β-galactosidase activities were similar. This study supports the idea that sex differences start very early in prenatal life and influence specific parameters, suggesting that it may be relevant to appreciate sex differences to cover knowledge gaps. This might lead to improving the diagnosis of risk prediction for pregnancy complications and achieving a more satisfactory monitoring of fetus health, even preventing future diseases in adulthood.

Project description:BackgroundUltrasound scanning during the 2nd or the 3rd trimester of pregnancy for fetal size disturbances screening is heavily dependent of the choice of the reference chart. This study aimed to assess the agreement of Salomon and the Intergrowth 21st equations in evaluating fetal biometric measurements in a rural area of Burkina Faso, and to measure the effect of changing a reference chart.MethodsData collected in Nazoanga, Burkina Faso, between October 2010 and October 2012, during a clinical trial evaluating the safety and efficacy of several antimalarial treatments in pregnant women were analyzed. We included singleton pregnancies at 16-36 weeks gestation as determined by ultrasound measurements of fetal bi-parietal diameter (BPD), head circumference (HC), abdominal circumference (AC) and femur length (FL). Expected mean and standard deviation at a given gestational age was computed using equations from Salomon references and using Intergrowth 21st standard. Then, z-scores were calculated and used subsequently to compare Salomon references with Intergrowth 21st standards.ResultsThe analysis included 276 singleton pregnancies. Agreement was poor except for HC: mean difference - 0.01, limits of agreement - 0.60 and 0.59. When AC was used as a surrogate of fetal size, switching from the reference of Salomon to the standards of Intergrowth 21st increased ten times the proportion of fetuses above the 90th percentile: 2.9 and 31.2%, respectively. Mean differences were larger in the third trimester than in the second trimester. However, agreement remained good for HC in both trimesters. Difference in the proportion of AC measurements above the 90th percentile using Salomon and Intergrowth 21st equations was greater in the second trimester (2.6 and 36.3%, respectively) than in the third trimester (3.5 and 19.8%, respectively). The greatest difference between the two charts was observed in the number of FL measurements classified as large in the second trimester (6.8 and 54.2%, using Salomon and Intergrowth 21st equations, respectively).ConclusionThe agreement between Intergrowth 21st and Salomon equations is poor apart from HC. This would imply different clinical decision regarding the management of the pregnancy.

Dataset Information

Fetal biometry and amniotic fluid volume assessment end-to-end automation using Deep Learning.

Publications

Fetal biometry and amniotic fluid volume assessment end-to-end automation using Deep Learning.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets