394 Brain Imaging Using a Novel Three-Dimensional Ultrasound System

Abstract

Diagnosis of hydrocephalus poses clinical challenges. Head computed tomography (CT) exposes the patient to carcinogenic ionizing radiation, at high cost. Magnetic resonance imaging (MRI) avoids ionizing radiation but has high cost, may require sedation, and requires reprogramming of electronic ventricular shunts affected by magnetic fields. In infants with open fontanelles, two dimensional (2D) ultrasound (US) is a radiation-free and low cost alternative to CT and MRI, but provides less visualization of the three-dimensional (3D) structure of brain and ventricles. We sought to determine the feasibility and accuracy of measurements of the brain and ventricles in an infant with hydrocephalus using a novel 3D US system, compared with conventional 2D US and MRI obtained during the clinical care of the patient. A micro-electro-mechanical systems chip with a gyroscope and accelerometer was externally affixed to a standard 2D US probe, allowing recording of probe orientation during imaging. Software developed by the research team captured 2D US images and plotted these into 3D space using their associated orientation data to create a 3D image volume. A 7-month-old infant with stable un-shunted hydrocephalus was enrolled following parental informed consent, under a protocol approved by the local institutional review board. The research device was paired with a 2D US system (Sonosite M-Turbo, p21 phased array transducer). Four 3D US scans were performed by emergency physicians with 2D source image depth 13cm, screen pixel resolution 490 width by 330 depth. Research 3D US volumes were compared with 2D US and MRI routinely obtained during the patient’s clinical care. Two researchers independently measured the 2D US, MR, and 3D US images using multiplanar and 3D volume rendering in open source software (3D Slicer). For 3D US, mean source image acquisition time was 10.49 seconds (range 5.15-14.3). Mean 3D reconstruction time was 52.36 seconds (range 29.67-66.97). A mean of 365 2D US frames were captured to construct the 3D volume (range 184-484). 2D US obtained during the patient’s clinical care recorded 57 2D frames (26 coronal and 31 sagittal images) over 272 seconds. Clinical 2D US image depth and resolution varied. MRI produced 24 axial images and 20 sagittal images, each with 256 by 256 pixel resolution. Mean measurements of brain and ventricles by the two observers are shown in the table. Example images are shown in the figure. Differences in measurements may be due to growth in the patient in the interval from 2D US until 3D ultrasound. 2D US occurred 1 month after MRI, and 3D US occurred approximately 2 months after MRI. Point-of-care 3D US allowed rapid acquisition of brain images, with linear and volume measurements comparable to those obtained by radiology-performed 2D US and MRI, two reference standards. Future research will apply this technique to a larger series of patients to better determine the diagnostic characteristics of this new technique.Tabled 1ModalityLateral ventricles, maximum medial-lateral distance (mm)Lateral ventricles, maximum cranial-caudal distance (mm)Lateral ventricles, maximum anterior-posterior distance (mm)Biparietal diameter, inner table to inner table calvarium (mm)Lateral ventricles volume (ml)MRI21.521.851.588.342.02D US24.023.154.790.7NA3D US23.822.353.297.740.9 Open table in a new tab