Automated 3D ultrasound image analysis for first trimester assessment of fetal health

Abstract

The first trimester fetal ultrasound scan is important to confirm fetal viability, to estimate the gestational age of the fetus, and to detect fetal anomalies early in pregnancy. First trimester ultrasound images have a different appearance than for the second trimester scan, reflecting the different stage of fetal development. There is limited literature on automation of image-based assessment for this earlier trimester, and most of the literature is focused on one specific fetal anatomy. In this paper, we consider automation to support first trimester fetal assessment of multiple fetal anatomies including both visualization and the measurements from a single 3-D ultrasound scan. We present a deep learning and image processing solution i) to perform semantic segmentation of the whole fetus, ii) to estimate plane orientation for standard biometry views, iii) to localize and automatically estimate biometry, and iv) to detect fetal limbs from a 3-D first trimester volume. Computational analysis methods were built using a real-world dataset (n=44 volumes). An evaluation on a further independent clinical dataset (n=21 volumes) showed that the automated methods approached human expert assessment of a 3D volume.

A multiple‐channel and atrous convolution network for ultrasound image segmentation

“…We attempt HC segmentation using images (800 × 540 pixels) from a medical image segmentation challenge. 34 We crop the images to 512 × 512. The datasets contain 788 fetal ultrasound images that are randomly divided into 718 images for training and 70 images for testing.…”

“…Therefore, accurate segmenting of relevant anatomical structures is the first step in measuring the head circumference (HC). [31][32][33][34] Lymph nodes are the most common metastatic site for cancer, and nodal status is critical for treatment decisions and prognosis. Analysis of lymph node features such as shape and size can supply valuable information for clinical diagnosis.…”

Right or wrong? Looking through the retrospectoscope to analyse predictions made a decade ago in prenatal diagnosis and fetal surgery

“…24 Artificial intelligence in the near future may overcome this inherent limitation of prenatal ultrasonography. 25,26 Furthermore the detection of fetal anomalies may move even earlier with advances in transvaginal imaging and operator experience 27 especially as novel imaging technologies are facilitating excellent post-mortem examination even in very small fetuses. 28 One might have expected that the place of fetal ultrasound in prenatal diagnosis will be diminished with the rapid advances being made with molecular genetic prenatal diagnosis.…”

“…Evidence points to the importance of experience (volume of ultrasound screens performed) in determining the skills of the sonographer 24 . Artificial intelligence in the near future may overcome this inherent limitation of prenatal ultrasonography 25,26 . Furthermore the detection of fetal anomalies may move even earlier with advances in transvaginal imaging and operator experience 27 especially as novel imaging technologies are facilitating excellent post‐mortem examination even in very small fetuses 28 .…”

Technology trends and applications of deep learning in ultrasonography: image quality enhancement, diagnostic support, and improving workflow efficiency

“…This new trend is another jump from the traditional ultrasound technology inspired by information technology (IT) and consumer electronics technology. The new AI-based applications range from enhancing ultrasound images [3][4][5][6][7][8] to smart and efficient improvement of workflow of healthcare business [9][10][11][12][13][14][15][16][17][18]. Deep learning applied to imaging chain has shown a big improvement in the efficiency and effect of processing, such as signal acquisition, adaptive beamforming, clutter suppression or compressive encodings for color Doppler [19].…”