Fast and Accurate U-Net Model for Fetal Ultrasound Image Segmentation

2022 ◽  
pp. 016173462110698
Author(s):  
Vahid Ashkani Chenarlogh ◽  
Mostafa Ghelich Oghli ◽  
Ali Shabanzadeh ◽  
Nasim Sirjani ◽  
Ardavan Akhavan ◽  
...  

U-Net based algorithms, due to their complex computations, include limitations when they are used in clinical devices. In this paper, we addressed this problem through a novel U-Net based architecture that called fast and accurate U-Net for medical image segmentation task. The proposed fast and accurate U-Net model contains four tuned 2D-convolutional, 2D-transposed convolutional, and batch normalization layers as its main layers. There are four blocks in the encoder-decoder path. The results of our proposed architecture were evaluated using a prepared dataset for head circumference and abdominal circumference segmentation tasks, and a public dataset (HC18-Grand challenge dataset) for fetal head circumference measurement. The proposed fast network significantly improved the processing time in comparison with U-Net, dilated U-Net, R2U-Net, attention U-Net, and MFP U-Net. It took 0.47 seconds for segmenting a fetal abdominal image. In addition, over the prepared dataset using the proposed accurate model, Dice and Jaccard coefficients were 97.62% and 95.43% for fetal head segmentation, 95.07%, and 91.99% for fetal abdominal segmentation. Moreover, we have obtained the Dice and Jaccard coefficients of 97.45% and 95.00% using the public HC18-Grand challenge dataset. Based on the obtained results, we have concluded that a fine-tuned and a simple well-structured model used in clinical devices can outperform complex models.

2014 ◽  
Vol 33 (4) ◽  
pp. 797-813 ◽  
Author(s):  
Sylvia Rueda ◽  
Sana Fathima ◽  
Caroline L. Knight ◽  
Mohammad Yaqub ◽  
Aris T. Papageorghiou ◽  
...  

2005 ◽  
Vol 58 (11-12) ◽  
pp. 548-552 ◽  
Author(s):  
Ljiljana Mladenovic-Segedi ◽  
Dimitrije Segedi

Introduction Former investigations have shown that the accuracy of fetal weight estimation is significantly higher if several ultrasonic fe?tal parameters are measured, because the total body mass depends on the size of fetal head, abdominal circumference and femur length. The aim of this investigation was to establish the best regression model, that is a number of combinations of fetal parameters providing the most accurate fetal weight estimation in utero in our population. Material and methods This prospective study was carried out at the Gynecology and Obstetrics Clinic of the Clinical Center Novi Sad. It included 270 pregnant women with singleton pregnancies within 72 hours of delivery who underwent ultrasound measurements of the biparietal diameter (BPD), head circumference (HC), ab?dominal circumference (AC) and femur length (FL). Results In regard to fetal weight estimation formulas, the deviation was lowest using regression models that simultaneously analyzed four fetal parameters (0.55%) with SD ?7.61%. In these models the estimates of fetal weights were within ?5% of actual birth weight in 48.89%, and within ?10% of actual birth weight in 81.48%. Good results were also obtained using AC, FL measurements (0.92% ? 8.20) as well as using AC, HC, FL measurements (-1.45% ? 7.81). In our sample the combination of AC and FL model gave better results in fetal weight estimation (0.92 ? 8.20%) than the one using BPD and AC (2.97 ? 8.83%). Furthermore, the model using parameters AC, HC and FL showed a lower error in accuracy (-1.45 ? 7.81%) than the model using BPD, AC and FL (2.51 ? 7.82%). Conclusion This investigation has confirmed that the accuracy of fetal weight estimation increases with the number of measured ultra?sonic fetal parameters. In our population the greatest accuracy was obtained using BPD, HC, AC and FL model. In cases when fast estimation of fetal weight is needed, AC, HC, FL model may be appropriate, but if fetal head circumference cannot be measured (amnion rupture and/or fetal head already in the pelvis) the AC, FL model should be used.


Open Medicine ◽  
2018 ◽  
Vol 13 (1) ◽  
pp. 374-383 ◽  
Author(s):  
Ismail Yaqub Maolood ◽  
Yahya Eneid Abdulridha Al-Salhi ◽  
Songfeng Lu

AbstractIn this study, an effective means for detecting cancer region through different types of medical image segmentation are presented and explained. We proposed a new method for cancer segmentation on the basis of fuzzy entropy with a level set (FELs) thresholding. The proposed method was successfully utilized to segment cancer images and then efficiently performed the segmentation of test ultrasound image, brain MRI, and dermoscopy image compared with algorithms proposed in previous studies. Results showed an excellent performance of the proposed method in detecting cancer image segmentation in terms of accuracy, precision, specificity, and sensitivity measures.


2012 ◽  
Vol 2012 ◽  
pp. 1-3 ◽  
Author(s):  
Khalil Al Tawil ◽  
Walid Salhi ◽  
Safiah Sultan ◽  
Mohammad Namshan ◽  
Saeed Mohammed

Meconium peritonitis pseudo-cyst is very rare. Its perinatal management is controversial and can be associated with increased fetomaternal morbidity and mortality. A 34-week gestation infant with large meconium peritonitis pseudo-cyst, detected by intrapartum fetal ultrasound study, had abnormally increased ratio of fetal abdominal circumference to head circumference. Intrapartum aspiration of the pseudo-cyst was performed and was followed by a smooth vaginal delivery. The postnatal course of the baby showed that early laparotomy was essential for stabilisation of the infant's general condition.


2019 ◽  
Vol 10 (1) ◽  
Author(s):  
Sander Galjaard ◽  
Lieveke Ameye ◽  
Christoph C. Lees ◽  
Anne Pexsters ◽  
Tom Bourne ◽  
...  

Abstract Background According to the WHO Multicentre Growth Reference Study Group recommendations, boys and girls have different growth trajectories after birth. Our aim was to develop gender-specific fetal growth curves in a low-risk population and to compare immediate birth outcomes. Methods First, second, and third trimester fetal ultrasound examinations were conducted between 2002 and 2012. The data was selected using the following criteria: routine examinations in uncomplicated singleton pregnancies, Caucasian ethnicity, and confirmation of gestational age by a crown-rump length (CRL) measurement in the first trimester. Generalized Additive Model for Location, Scale and Shape (GAMLSS) was used to align the time frames of the longitudinal fetal measurements, corresponding with the methods of the postnatal growth curves of the WHO MGRS Group. Results A total of 27,680 complete scans were selected from the astraia© ultrasound database representing 12,368 pregnancies. Gender-specific fetal growth curves for biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC), and femur length (FL) were derived. The HC and BPD were significantly larger in boys compared to girls from 20 weeks of gestation onwards (p < 0.001) equating to a 3-day difference at 20–24 weeks. Boys were significantly heavier, longer, and had greater head circumference than girls (p < 0.001) at birth. The Apgar score at 1 min (p = 0.01) and arterial cord pH (p < 0.001) were lower in boys. Conclusions These longitudinal fetal growth curves for the first time allow integration with neonatal and pediatric WHO gender-specific growth curves. Boys exceed head growth halfway of the pregnancy, and immediate birth outcomes are worse in boys than girls. Gender difference in intrauterine growth is sufficiently distinct to have a clinically important effect on fetal weight estimation but also on the second trimester dating. Therefore, these differences might already play a role in early fetal or immediate neonatal management.


Sign in / Sign up

Export Citation Format

Share Document