Automatic Detection and Measurement of Structures in Fetal Head Ultrasound Volumes Using Sequential Estimation and Integrated Detection Network (IDN)

Abstract

Routine ultrasound exam in the second and third trimesters of pregnancy involves manually measuring fetal head and brain structures in 2D scans. The procedure requires a sonographer to find the standardized visualization planes with a probe and manually place measurement calipers on the structures of interest. The process is tedious, time consuming, and introduces user variability into the measurements. This paper proposes an Automatic Fetal Head and Brain (AFHB) system for automatically measuring anatomical structures from 3D ultrasound volumes. The system searches the 3D volume in a hierarchy of resolutions and by focusing on regions that are likely to be the measured anatomy. The output is a standardized visualization of the plane with correct orientation and centering as well as the biometric measurement of the anatomy. The system is based on a novel framework for detecting multiple structures in 3D volumes. Since a joint model is difficult to obtain in most practical situations, the structures are detected in a sequence, one-byone. The detection relies on Sequential Estimation techniques, frequently applied to visual tracking. The interdependence of structure poses and strong prior information embedded in our domain yields faster and more accurate results than detecting the objects individually. The posterior distribution of the structure pose is approximated at each step by sequential Monte Carlo. The samples are propagated within the sequence across multiple structures and hierarchical levels. The probabilistic model helps solve many challenges present in the ultrasound images of the fetus such as speckle noise, signal drop-out, shadows caused by bones, and appearance variations caused by the differences in the fetus gestational age. This is possible by discriminative learning on an extensive database of scans comprising more than two thousand volumes and more than thirteen thousand annotations. The average difference between ground truth and automatic measu- ements is below 2 mm with a running time of 6.9 seconds (GPU) or 14.7 seconds (CPU). The accuracy of the AFHB system is within inter-user variability and the running time is fast, which meets the requirements for clinical use.

Results

Figure 1: The automatic measurement results (cyan) compared to ground truth (red) for Cerebellum (CER), Cisterna Magna (CM), and Lateral Ventricles (LV). The row of each structure shows results with approximately average plane error. Plane and length errors on the left are reported (in mm). Note, that the cases with higher error are still acceptable for clinical use. The last two columns show the agreement of the detection plane in the sagittal and coronal cross section.

CER: 4.89, 0.88
CM: 0.57, 0.07
LV: 0.64, 0.25

Figure 2: The automatic measurement results (cyan) compared to ground truth (red) for Occipitofrontal Diameter (OFD), Biparietal Diameter (BPD), and Head Circumference (HC). Plane and length errors on the left are reported (in mm).

OFD: 2.83, 0.77
BPD: 2.90, 1.09
HC: 10.93, 5.86

Figure 3: The automatic detection results (cyan) compared to ground truth (red) for Corpus Callosum (CC) and Choroid Plexus (CP). The errors of detecting center and plane orientation are indicated (in mm). Annotation lines are shown for reference but they are not used clinically.

CC: 0.22, 3.17
CP: 0.88, 3.92

Publications and Further Reading