Ultrasound is one of the most commonly used imaging methodologies in obstetrics to monitor the growth of a fetus during the gestation period. Specifically, ultrasound images are routinely utilized to gather fetal information, including body measurements, anatomy structure, fetal movements, and pregnancy complications. Recent developments in artificial intelligence and computer vision provide new methods for the automated analysis of medical images in many domains, including ultrasound images. We present a full end-to-end framework for segmenting, measuring, and estimating fetal gestational age and weight based on two-dimensional ultrasound images of the fetal head. Our segmentation framework is based on the following components: (i) eight segmentation architectures (UNet, UNet Plus, Attention UNet, UNet 3+, TransUNet, FPN, LinkNet, and Deeplabv3) were fine-tuned using lightweight network EffientNetB0, and (ii) a weighted voting method for building an optimized ensemble transfer learning model (ETLM). On top of that, ETLM was used to segment the fetal head and to perform analytic and accurate measurements of circumference and seven other values of the fetal head, which we incorporated into a multiple regression model for predicting the week of gestational age and the estimated fetal weight (EFW). We finally validated the regression model by comparing our result with expert physician and longitudinal references. We evaluated the performance of our framework on the public domain dataset HC18: we obtained 98.53% mean intersection over union (mIoU) as the segmentation accuracy, overcoming the state-of-the-art methods; as measurement accuracy, we obtained a 1.87 mm mean absolute difference (MAD). Finally we obtained a 0.03% mean square error (MSE) in predicting the week of gestational age and 0.05% MSE in predicting EFW.

In ultrasound images, the speckle noise phenomenon is a typical issue that limits the accuracy of the images obtained from any quantitative measurement. Noise elimination in ultrasound images in the pre-processing stage is an important step to make the image fit for the following steps involved in image segmentation. However, various noises produce a range of significant impacts on the quality of the image and thus tend to affect the interpretation of neural networks. In this study, 35 different image processing techniques, including denoising, feature descriptor, edge detection, miscellaneous, clustering, morphological, sharpening, and composite techniques were used to determine the most optimal ultrasound image in order to improve the segmentation of ultrasound images. The performance of image processing techniques was compared using a statistical parameter, peak signal-to-noise ratio (PSNR). All techniques were examined on deep learning segmentation approach. The segmentation performance was evaluated using mean intersection over union (mIoU) and mean pixel accuracy (mPA). The results showed that the composite image technique (FancyPCA + Bilateral + Gabor) slightly enhanced the segmentation performance, with the most ideal mIoU 0.96893%, mPA 0.97831%, and average PSNR 53.034 dB.