Automatic Detection and Measurement of Fetal Biparietal Diameter and Femur Length —Feasibility on a Portable Ultrasound Device

Abstract

An automatic method able to recognize a presented section through the biparietal plane of the fetal head and a section through the fetal femur in ultrasound images is developed. Once the correct anatomical section for measurement is identified by the machine, the placement of the measurement calipers is automatically determined by fitting an active contour model to the structure of interest. The fetal biparietal diameter (BPD) and femur length (FL) are then measured automatically. The validation data set contained 167 and 197 Bmode images for BPD and FL measurements, respectively. The images were acquired using 4 different ultrasound scanners, which resulted in varied image quality and gain settings. The mean gestational age (GA) of the fetuses was 19.4 weeks, range 16 to 41 weeks. A measurement success rate of 90% was achieved for both BPD and FL. The correlation coefficients between the manual and automatic measurements were 0.995 (BPD) and 0.967 (FL), mean errors were 0.5 mm (BPD) and -1.7 mm (FL) and error range with 95% confidence interval (CI) were −3.8 -4.8 mm (BPD) and −11.4 -8.1 mm (FL). The automatic measurement results were consistent in both high and low gain settings. The intraclass correlation coefficients between manual and automatic measurements were 0.995 (95% CI; 0.981 -0.999) for BPD in high gain, 1.0 (95% CI; 0.998 -1.0) for BPD in low gain, 0.998 (95% CI; 0.991 -0.999) for FL in high gain and 0.999 (95% CI; 0.996 -1.0) for FL in low gain settings. The method was implemented on a prototype, portable ultrasound machine designed to be used in low-and middle-income countries (LMIC). The overall performance of the method supports our hypothesis that automated methods can be used and are beneficial in a clinical setting.

Automatic linear measurements of the fetal brain on MRI with deep neural networks

“…To summarize, while biometric linear measurements of the fetal brain are an essential part of fetal development assessment, they are currently performed manually. While automatic methods for the computation of US-based biometric linear measurements are available, e.g., biparietal diameter [21,22], fetal head circumference [23] and femur length [21], no such methods are available for fetal MRI.…”

Automatic Measurement of Fetal Abdominal Biometric Parameters—Feasibility on a Portable Ultrasound Device

“…This device is aimed to be used by trained ultrasound operators and midwives trained, or undergoing training, in ultrasound in LMIC. As a follow-up of our prior work [22] [23] on automatizing BPD, FL and fetal abdominal measurements, the aim of this study is to develop a fully automatic method for detecting and measuring the fetal abdominal section in a B-mode ultrasound image, and for measuring the MAD or AC. The method is designed to be run on a low cost, easy-to-use portable ultrasound machine suited for use in LMIC.…”

AutoFB: Automating Fetal Biometry Estimation from Standard Ultrasound Planes

“…There is extensive work on segmentation of anatomical structures in standard US planes, specifically those concerning second and third trimester screening [19]. These techniques can support automated fetal biometry, including measurements on the head [24,13,16,23,15,4], femur [15,12], and abdominal section [14]. These methods, however, rely on prior knowledge of which measurement to perform on a given image.…”

“…A fully automated biometry system should both identify which standard plane is being imaged and whether it is of sufficient quality to perform the relevant measurements. Automatic image quality assessment has been investigated, including adequate magnification, symmetry and the visibility of relevant anatomical structures within the image [17,15]. Such methods together with classification of standard planes [1] can be used to extract appropriate planes for fetal biometry from US video or image collections [9].…”