Towards improved ultrasound‐based analysis and 3D visualization of the fetal brain using the 3D Slicer

Magnetic resonance imaging (MRI) provides useful three-dimensional (3D) information; however, there are some restrictions on its use during pregnancy due to safety concerns. In addition, fetal movements can create artifacts on MR images, as image quality depends on position of the fetus and placenta. In the past decade, 3D ultrasound imaging has been used in clinical practice to investigate the formation and volumetric size of critical anatomical structures of the fetus. However, current techniques rely mainly on analysis of sections of interest that do not integrate anatomical information concerning the shape of these structures. We provide a brief description of a workflow for semi-automated segmentation and 3D visualization of fetal ultrasound volumes in the second trimester using the 3D Slicer open source software1,2. Our workflow allowed quantitative image analysis of the choroid plexus and cerebrum from 3D ultrasound images. We acquired 3D ultrasound volumes from five healthy pregnant women at 12 (n=2), 14 (n=2) and 19 (n=1) weeks of gestation. Informed consent was obtained in each case. We used a Voluson E6 (GE Medical Systems, Zipf, Austria) ultrasound machine with a RAB4-8-D/OB D/4D 8-MHz transabdominal transducer. Our workflow consisted of four steps (Figure 1). Firstly, we imported DICOM (digital imaging and communications in medicine) ultrasound volumes into the 3D Slicer. We then used the ‘Grow Cut Segmentation’ algorithm3 of the interactive Editor module to extract critical structures from the ultrasound volumes. We reconstructed 3D surface models from segmented regions using the ‘Marching Cubes’ algorithm4, and finally computed the volume of 3D anatomical models using the ‘Label Statistics’ module of the software. Figure 1 Flowchart describing the 3D Slicer workflow used in this study. DICOM, digital imaging and communications in medicine. Figure 2 shows a 3D surface model of the choroid plexus and cerebrum reconstructed from the original 3D ultrasound volumes. The corresponding volumes of these structures at 12, 14 and 19 weeks’ gestation were, respectively: 431.1mm3, 698.9mm3 and 1203.3 mm3 for the choroid plexus and 183.6 mm3, 282.8 mm3 and 469.8 mm3 for the cerebrum. Figure 2 Result of ‘Grow Cut Segmentation’ of the fetal brain using the 3D Slicer. The blue structure represents the choroid plexus, and the yellow structure the cerebrum at 14 weeks of gestation (axial and coronal views). Using the 3D Slicer, we were able to obtain patient-specific quantitative information and 3D visualization of anatomical structures within the fetal brain. We anticipate being able to perform segmentation that accurately matches the anatomy using different methods. We believe this method, combined with ultrasound or MRI data, will be helpful in monitoring fetal development and detecting anomalies of the brain as well as other anatomical structures.

[1]  Ron Kikinis,et al.  3D Slicer , 2012, 2004 2nd IEEE International Symposium on Biomedical Imaging: Nano to Macro (IEEE Cat No. 04EX821).

[2]  William E. Lorensen,et al.  Marching cubes: A high resolution 3D surface construction algorithm , 1987, SIGGRAPH.